@misc{schulz_daidalos_2026,
	title = {Daidalos {Project}: {Development} of an {Infrastructure} for the {Use} of {Natural} {Language} {Processing} for {Researchers} in {Classical} {Philology}},
	shorttitle = {Daidalos {Project}},
	url = {https://zenodo.org/records/19064996},
	abstract = {The presentation introduces the Daidalos Project, a German research initiative based at Humboldt University Berlin that develops infrastructure to support the use of natural language processing (NLP) in classical philology. Running from 2023 to 2026 (with a possible extension to 2030), the project is funded by the German Research Foundation and brings together experts from classical philology, corpus linguistics, and information technology. Rather than conducting original research, its primary goal is to make advanced computational methods more accessible to scholars working with Latin and Ancient Greek texts.

Daidalos provides a wide range of services, including NLP servers with public APIs, a user-friendly JupyterLite environment, no-code graphical interfaces, and extensive documentation. It also offers educational materials such as Jupyter notebooks, open-access teaching resources, and curated datasets. Social initiatives — like workshops, conferences, and research tandems — play a central role in engaging the academic community.

Workshops combine theoretical introductions with hands-on sessions, while research tandems pair scholars with project members to collaboratively design and prototype NLP-based studies. Results and workflows are openly published, reflecting the project’s commitment to open science and sustainable research support.},
	language = {eng},
	urldate = {2026-03-17},
	author = {Schulz, Konstantin},
	month = mar,
	year = {2026},
	doi = {10.5281/zenodo.19064996},
	keywords = {Classics, Information infrastructure, Infrastructure, Natural Language Processing, Natural language processing},
}

The presentation introduces the Daidalos Project, a German research initiative based at Humboldt University Berlin that develops infrastructure to support the use of natural language processing (NLP) in classical philology. Running from 2023 to 2026 (with a possible extension to 2030), the project is funded by the German Research Foundation and brings together experts from classical philology, corpus linguistics, and information technology. Rather than conducting original research, its primary goal is to make advanced computational methods more accessible to scholars working with Latin and Ancient Greek texts. Daidalos provides a wide range of services, including NLP servers with public APIs, a user-friendly JupyterLite environment, no-code graphical interfaces, and extensive documentation. It also offers educational materials such as Jupyter notebooks, open-access teaching resources, and curated datasets. Social initiatives — like workshops, conferences, and research tandems — play a central role in engaging the academic community. Workshops combine theoretical introductions with hands-on sessions, while research tandems pair scholars with project members to collaboratively design and prototype NLP-based studies. Results and workflows are openly published, reflecting the project’s commitment to open science and sustainable research support.

@misc{cobb_module_2026,
	type = {Open {Educational} {Resource}},
	title = {Module 3: {LLMs} and {Digital} {Classics}},
	shorttitle = {Module 3},
	url = {https://zenodo.org/records/19056407},
	doi = {10.5281/zenodo.19056407},
	abstract = {Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/). 
Die OER-Ressource wird in zwei Fassungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszeit beträgt knapp 3h. Für weitere Informationen s. readme.txt.
Inhalt: 
Was ist ein 'Large Language Model'?
Wie können mir LLMs beim Studium helfen?
Chatbots
Risiken und Herausforderungen
Anwendung von LLMs
----
The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/).
The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 3 hours to complete. For further information, see readme.txt.
Content:
What is a Large Language Model?
How do LLMs relate to my studies?
Chatbots
Risks and Challenges
Applying LLMs},
	language = {en, de},
	urldate = {2026-03-16},
	publisher = {Zenodo},
	author = {Cobb, George},
	collaborator = {Beyer, Andrea},
	month = mar,
	year = {2026},
	keywords = {Digital Classics, Klassische Philologie, LLM, Large Language Models, Latin Philology, Moodle},
}

Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/).  Die OER-Ressource wird in zwei Fassungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszeit beträgt knapp 3h. Für weitere Informationen s. readme.txt. Inhalt:  Was ist ein 'Large Language Model'? Wie können mir LLMs beim Studium helfen? Chatbots Risiken und Herausforderungen Anwendung von LLMs —- The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/). The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 3 hours to complete. For further information, see readme.txt. Content: What is a Large Language Model? How do LLMs relate to my studies? Chatbots Risks and Challenges Applying LLMs

@misc{schulz_technical_2026,
	type = {Pitch},
	title = {Technical {Standards} for {Latin} and {Ancient} {Greek}: {Referencing}, {Evaluation}, {Documentation}},
	shorttitle = {Technical {Standards} for {Latin} and {Ancient} {Greek}},
	url = {https://zenodo.org/records/18863970},
	abstract = {This pitch argues that rapid advances in AI and NLP for Latin and Ancient Greek have outpaced the development and adoption of shared technical standards.

While the field already has relevant frameworks—such as CITE URNs, Canonical Text Services, model cards, datasheets, and platforms like Hugging Face—their implementation is uneven and insufficient for sustainable research.

The core question is how to make NLP research for ancient languages more sustainable. Proposed steps include:





comprehensive surveys of methods and performance,




identifying weaknesses in existing standards or their adoption,




developing clear guidelines and best practices,




fostering collaboration through workshops, benchmarking, and shared code.



Concrete “first issues” include standardizing lemmatization via canonical lemma lists, creating domain-specific annotation guidelines for named entity recognition, and agreeing on canonical encoding and normalization practices.

Overall, the presentation calls for coordinated community efforts to align innovation in digital classics with robust referencing, documentation, and evaluation standards.},
	language = {eng},
	urldate = {2026-03-04},
	author = {Schulz, Konstantin},
	month = mar,
	year = {2026},
	doi = {10.5281/zenodo.18863970},
	keywords = {Classics, International standardisation, Natural Language Processing, Natural language processing, Standardisation},
}

This pitch argues that rapid advances in AI and NLP for Latin and Ancient Greek have outpaced the development and adoption of shared technical standards. While the field already has relevant frameworks—such as CITE URNs, Canonical Text Services, model cards, datasheets, and platforms like Hugging Face—their implementation is uneven and insufficient for sustainable research. The core question is how to make NLP research for ancient languages more sustainable. Proposed steps include: comprehensive surveys of methods and performance, identifying weaknesses in existing standards or their adoption, developing clear guidelines and best practices, fostering collaboration through workshops, benchmarking, and shared code. Concrete “first issues” include standardizing lemmatization via canonical lemma lists, creating domain-specific annotation guidelines for named entity recognition, and agreeing on canonical encoding and normalization practices. Overall, the presentation calls for coordinated community efforts to align innovation in digital classics with robust referencing, documentation, and evaluation standards.

@misc{cobb_module_2026,
	type = {Open {Educational} {Resource}},
	title = {Module 2: {Introduction} to {Digital} {Editions} / {Einführung} in {Digitale} {Editionen}},
	shorttitle = {Module 2},
	url = {https://zenodo.org/records/18848616},
	doi = {10.5281/zenodo.18848616},
	abstract = {Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/). 
Die OER-Ressource wird in zwei Fassungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszeit beträgt knapp 2h. Für weitere Informationen s. readme.txt.
Inhalt: 
Was ist eine Digital Edition?
Digital Latin Library
DiogenesWeb
Einführung in XML
Vergils Aeneis und XML
----
The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/).
The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 2 hours to complete. For further information, see readme.txt.
Content:
What is a Digital Edition?
Digital Latin Library
DiogenesWeb
Introduction to XML
Virgil’s Aeneid \& XML},
	language = {en, de},
	urldate = {2026-03-03},
	publisher = {Zenodo},
	author = {Cobb, George},
	collaborator = {Beyer, Andrea},
	month = mar,
	year = {2026},
	keywords = {Digital Classics, Digital Editions, Klassische Philologie, Latin Philology, Moodle},
}

Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/).  Die OER-Ressource wird in zwei Fassungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszeit beträgt knapp 2h. Für weitere Informationen s. readme.txt. Inhalt:  Was ist eine Digital Edition? Digital Latin Library DiogenesWeb Einführung in XML Vergils Aeneis und XML —- The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/). The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 2 hours to complete. For further information, see readme.txt. Content: What is a Digital Edition? Digital Latin Library DiogenesWeb Introduction to XML Virgil’s Aeneid & XML

@misc{beyer_omniprasente_2026,
	address = {Tübingen},
	type = {Workshop},
	title = {Omnipräsente {KI}: {Autonomie} durch {Kompetenz} – {Warum} der {Erwerb} der alten {Sprachen} trotz {KI} nicht obsolet wird},
	doi = {https://doi.org/10.5281/zenodo.18664004},
	abstract = {This talk explores the enduring relevance of classical language acquisition in the age of artificial intelligence (AI), arguing that despite the growing capabilities of AI systems, proficiency in ancient languages remains essential for genuine scholarly autonomy and deep understanding. Drawing on the example of the "Passio Perpetuae et Felicitatis", the presentation examines how AI tools - while useful as tutors, assistants, and even for exam creation - cannot replace the critical, interpretive, and hermeneutic competencies developed through direct engagement with original texts. The analysis highlights the limitations of AI in grasping historical context, implicit world knowledge, and nuanced textual variations across different manuscript traditions among other things. It further discusses the risks of cognitive offloading, superficial learning, and the erosion of intellectual sovereignty when relying on AI without deep linguistic and methodological foundations. The talk concludes that true academic independence in theological and philological research requires not just a fair and open access to AI, but a solid command of the source languages, enabling scholars to critically evaluate, interpret, and contextualise texts beyond what AI can generate.
----
Dieser Vortrag untersucht die anhaltende Relevanz des Erwerbs klassischer Sprachen im Zeitalter der künstlichen Intelligenz (KI) und argumentiert, dass trotz der wachsenden Fähigkeiten von KI-Systemen die Beherrschung alter Sprachen für echte wissenschaftliche Autonomie und tiefes Verständnis nach wie vor unerlässlich ist. Anhand des Beispiels der „Passio Perpetuae et Felicitatis” untersucht der Vortrag, wie KI-Tools – obwohl sie als Tutoren, Assistenten und sogar für die Erstellung von Prüfungen nützlich sind – die kritischen, interpretativen und hermeneutischen Kompetenzen nicht ersetzen können, die durch die direkte Auseinandersetzung mit Originaltexten entwickelt werden. Die Analyse hebt die Grenzen der KI hervor, unter anderem beim Erfassen des historischen Kontexts, bei implizitem Weltwissen und nuancierten Textvariationen in verschiedenen Manuskripttraditionen. Darüber hinaus werden die Risiken des "cognitive offloading", des oberflächlichen Lernens und der Erosion der intellektuellen Souveränität diskutiert, wenn man sich ohne fundierte sprachliche und methodische Grundlagen auf KI verlässt. Der Vortrag kommt zu dem Schluss, dass echte akademische Unabhängigkeit in der theologischen und philologischen Forschung nicht nur einen fairen und offenen Zugang zu KI erfordert, sondern auch eine solide Beherrschung der Ausgangssprachen, die es Wissenschaftlern ermöglicht, Texte über das hinaus, was KI generieren kann, kritisch zu bewerten, zu interpretieren und zu kontextualisieren.},
	author = {Beyer, Andrea},
	month = dec,
	year = {2026},
}

This talk explores the enduring relevance of classical language acquisition in the age of artificial intelligence (AI), arguing that despite the growing capabilities of AI systems, proficiency in ancient languages remains essential for genuine scholarly autonomy and deep understanding. Drawing on the example of the "Passio Perpetuae et Felicitatis", the presentation examines how AI tools - while useful as tutors, assistants, and even for exam creation - cannot replace the critical, interpretive, and hermeneutic competencies developed through direct engagement with original texts. The analysis highlights the limitations of AI in grasping historical context, implicit world knowledge, and nuanced textual variations across different manuscript traditions among other things. It further discusses the risks of cognitive offloading, superficial learning, and the erosion of intellectual sovereignty when relying on AI without deep linguistic and methodological foundations. The talk concludes that true academic independence in theological and philological research requires not just a fair and open access to AI, but a solid command of the source languages, enabling scholars to critically evaluate, interpret, and contextualise texts beyond what AI can generate. —- Dieser Vortrag untersucht die anhaltende Relevanz des Erwerbs klassischer Sprachen im Zeitalter der künstlichen Intelligenz (KI) und argumentiert, dass trotz der wachsenden Fähigkeiten von KI-Systemen die Beherrschung alter Sprachen für echte wissenschaftliche Autonomie und tiefes Verständnis nach wie vor unerlässlich ist. Anhand des Beispiels der „Passio Perpetuae et Felicitatis” untersucht der Vortrag, wie KI-Tools – obwohl sie als Tutoren, Assistenten und sogar für die Erstellung von Prüfungen nützlich sind – die kritischen, interpretativen und hermeneutischen Kompetenzen nicht ersetzen können, die durch die direkte Auseinandersetzung mit Originaltexten entwickelt werden. Die Analyse hebt die Grenzen der KI hervor, unter anderem beim Erfassen des historischen Kontexts, bei implizitem Weltwissen und nuancierten Textvariationen in verschiedenen Manuskripttraditionen. Darüber hinaus werden die Risiken des "cognitive offloading", des oberflächlichen Lernens und der Erosion der intellektuellen Souveränität diskutiert, wenn man sich ohne fundierte sprachliche und methodische Grundlagen auf KI verlässt. Der Vortrag kommt zu dem Schluss, dass echte akademische Unabhängigkeit in der theologischen und philologischen Forschung nicht nur einen fairen und offenen Zugang zu KI erfordert, sondern auch eine solide Beherrschung der Ausgangssprachen, die es Wissenschaftlern ermöglicht, Texte über das hinaus, was KI generieren kann, kritisch zu bewerten, zu interpretieren und zu kontextualisieren.

@misc{liu_posterverse_2026,
	title = {{PosterVerse}: {A} {Full}-{Workflow} {Framework} for {Commercial}-{Grade} {Poster} {Generation} with {HTML}-{Based} {Scalable} {Typography}},
	shorttitle = {{PosterVerse}},
	url = {http://arxiv.org/abs/2601.03993},
	doi = {10.48550/arXiv.2601.03993},
	abstract = {Commercial-grade poster design demands the seamless integration of aesthetic appeal with precise, informative content delivery. Current automated poster generation systems face significant limitations, including incomplete design workflows, poor text rendering accuracy, and insufficient flexibility for commercial applications. To address these challenges, we propose PosterVerse, a full-workflow, commercial-grade poster generation method that seamlessly automates the entire design process while delivering high-density and scalable text rendering. PosterVerse replicates professional design through three key stages: (1) blueprint creation using fine-tuned LLMs to extract key design elements from user requirements, (2) graphical background generation via customized diffusion models to create visually appealing imagery, and (3) unified layout-text rendering with an MLLM-powered HTML engine to guarantee high text accuracy and flexible customization. In addition, we introduce PosterDNA, a commercial-grade, HTML-based dataset tailored for training and validating poster design models. To the best of our knowledge, PosterDNA is the first Chinese poster generation dataset to introduce HTML typography files, enabling scalable text rendering and fundamentally solving the challenges of rendering small and high-density text. Experimental results demonstrate that PosterVerse consistently produces commercial-grade posters with appealing visuals, accurate text alignment, and customizable layouts, making it a promising solution for automating commercial poster design. The code and model are available at https://github.com/wuhaer/PosterVerse.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Liu, Junle and Zhang, Peirong and Zhang, Yuyi and Yan, Pengyu and Zhou, Hui and Zhou, Xinyue and Guo, Fengjun and Jin, Lianwen},
	month = jan,
	year = {2026},
	note = {arXiv:2601.03993 [cs]},
	keywords = {Computer Science - Computer Vision and Pattern Recognition},
}

Commercial-grade poster design demands the seamless integration of aesthetic appeal with precise, informative content delivery. Current automated poster generation systems face significant limitations, including incomplete design workflows, poor text rendering accuracy, and insufficient flexibility for commercial applications. To address these challenges, we propose PosterVerse, a full-workflow, commercial-grade poster generation method that seamlessly automates the entire design process while delivering high-density and scalable text rendering. PosterVerse replicates professional design through three key stages: (1) blueprint creation using fine-tuned LLMs to extract key design elements from user requirements, (2) graphical background generation via customized diffusion models to create visually appealing imagery, and (3) unified layout-text rendering with an MLLM-powered HTML engine to guarantee high text accuracy and flexible customization. In addition, we introduce PosterDNA, a commercial-grade, HTML-based dataset tailored for training and validating poster design models. To the best of our knowledge, PosterDNA is the first Chinese poster generation dataset to introduce HTML typography files, enabling scalable text rendering and fundamentally solving the challenges of rendering small and high-density text. Experimental results demonstrate that PosterVerse consistently produces commercial-grade posters with appealing visuals, accurate text alignment, and customizable layouts, making it a promising solution for automating commercial poster design. The code and model are available at https://github.com/wuhaer/PosterVerse.

@unpublished{beyer_daidalos_2026,
	type = {In {Vorbereitung}},
	title = {Daidalos: {KI} in der {Klassischen} {Philologie} - {Historical} {Languages} and {AI}},
	language = {de},
	collaborator = {Beyer, Andrea and Schulz, Konstantin},
	year = {2026},
}

@incollection{beyer_textanalyse_2026,
	address = {Baden-Baden},
	series = {Paradeigmata},
	title = {Textanalyse mit {KI}: {Wie} kommt sie in die {Klassische} {Philologie}?},
	abstract = {Die Einführung von Künstlicher Intelligenz (KI) in die Klassische Philologie eröffnet neue Möglichkeiten für die systematische Textanalyse, kombiniert qualitative und quantitative Ansätze und erfordert eine fundierte methodische Kompetenz. Der Beitrag untersucht die notwendigen digitalen, daten- und KI-Kompetenzen (Digital Literacies) für Forschung, Lehre und Unterricht, basierend auf drei empirisch fundierten Fallbeispielen: einer Forschungskooperation zur Textanalyse mit Word Embeddings, einer Lehrveranstaltung zur Sentiment-Analyse im Kontext der Plinius-Briefe sowie einer modellhaften Unterrichtseinheit zum reflektierten Umgang mit KI-Chatbots. Die Darstellung zeigt, dass der erfolgreiche Einsatz von KI-Tools in der Klassischen Philologie nicht allein von technischen Fähigkeiten abhängt, sondern eine enge Verzahnung von fachlicher Expertise, methodischer Reflexion und digitaler Kompetenz erfordert. Insbesondere die Triangulation qualitativer und quantitativer Methoden sowie die kritische Auseinandersetzung mit den Grenzen und Vorurteilen von KI-Modellen sind entscheidend. Der Beitrag plädiert für eine fach- und fallspezifische KI-Bildung, die durch interdisziplinäre Kooperationen, Communities of Practice und forschungsorientierte Lehr-Lernformate gestärkt werden muss, um die Klassische Philologie zukunftsfähig zu machen.},
	language = {Deutsch},
	booktitle = {Digitalität im {Unterricht} der {Alten} {Sprachen}},
	publisher = {Nomos-Verlag},
	author = {Beyer, Andrea},
	editor = {Faller, Stefan and Polleichtner, Wolfgang},
	year = {2026},
	keywords = {bold, scientific},
	pages = {im Druck},
}

Die Einführung von Künstlicher Intelligenz (KI) in die Klassische Philologie eröffnet neue Möglichkeiten für die systematische Textanalyse, kombiniert qualitative und quantitative Ansätze und erfordert eine fundierte methodische Kompetenz. Der Beitrag untersucht die notwendigen digitalen, daten- und KI-Kompetenzen (Digital Literacies) für Forschung, Lehre und Unterricht, basierend auf drei empirisch fundierten Fallbeispielen: einer Forschungskooperation zur Textanalyse mit Word Embeddings, einer Lehrveranstaltung zur Sentiment-Analyse im Kontext der Plinius-Briefe sowie einer modellhaften Unterrichtseinheit zum reflektierten Umgang mit KI-Chatbots. Die Darstellung zeigt, dass der erfolgreiche Einsatz von KI-Tools in der Klassischen Philologie nicht allein von technischen Fähigkeiten abhängt, sondern eine enge Verzahnung von fachlicher Expertise, methodischer Reflexion und digitaler Kompetenz erfordert. Insbesondere die Triangulation qualitativer und quantitativer Methoden sowie die kritische Auseinandersetzung mit den Grenzen und Vorurteilen von KI-Modellen sind entscheidend. Der Beitrag plädiert für eine fach- und fallspezifische KI-Bildung, die durch interdisziplinäre Kooperationen, Communities of Practice und forschungsorientierte Lehr-Lernformate gestärkt werden muss, um die Klassische Philologie zukunftsfähig zu machen.

@incollection{beyer_daidalos_2026,
	series = {Digital {Classics} {Online}},
	title = {Daidalos: {NER} for {Literary} {Studies} on {Latin} and {Ancient} {Greek} {Texts}},
	abstract = {Literary texts offer a wealth of unstructured data that can be harnessed for data-driven text analysis through natural language processing (NLP). Named Entity Recognition and Classification (NER) is a crucial initial step in this process, enabling the automatic identification of entities such as persons, organizations, locations, and dates. However, NER faces significant challenges, particularly with his-torical texts in low-resource languages like Latin and Ancient Greek, due to limited annotated corpora and the dynamic nature of language. This paper explores the evolution of NER from simple extraction to semantics-aware entity disambiguation and linking, highlighting the importance of multi-layer anno-tation systems to enhance data quality and model accuracy. The interdisciplinary Daidalos project aims to bridge the gap between Digital Humanities and Classical Studies by providing an NLP infra-structure that supports various data-driven research methods, among others NER. One of the pro-ject's case studies demonstrates the potential of NER in classical literary studies; this is accompanied by proposals on other NER related literary research questions, e.g. on authorship attribution and ste-reotyping. Additionally, the paper offers some thoughts about teaching NER, presenting a framework to assess the required level of Digital Literacies when working on a specific research question. Finally, it discusses the implications of generative AI and Large Language Models (LLM) on NER and NLP in Classics, emphasizing the challenges for independent research posed by the high costs and limited transparency of LLMs.},
	booktitle = {Nomina {Omina}},
	author = {Beyer, Andrea},
	editor = {Berti, Monica},
	year = {2026},
	keywords = {bold, scientific},
	pages = {im Druck},
}

Literary texts offer a wealth of unstructured data that can be harnessed for data-driven text analysis through natural language processing (NLP). Named Entity Recognition and Classification (NER) is a crucial initial step in this process, enabling the automatic identification of entities such as persons, organizations, locations, and dates. However, NER faces significant challenges, particularly with his-torical texts in low-resource languages like Latin and Ancient Greek, due to limited annotated corpora and the dynamic nature of language. This paper explores the evolution of NER from simple extraction to semantics-aware entity disambiguation and linking, highlighting the importance of multi-layer anno-tation systems to enhance data quality and model accuracy. The interdisciplinary Daidalos project aims to bridge the gap between Digital Humanities and Classical Studies by providing an NLP infra-structure that supports various data-driven research methods, among others NER. One of the pro-ject's case studies demonstrates the potential of NER in classical literary studies; this is accompanied by proposals on other NER related literary research questions, e.g. on authorship attribution and ste-reotyping. Additionally, the paper offers some thoughts about teaching NER, presenting a framework to assess the required level of Digital Literacies when working on a specific research question. Finally, it discusses the implications of generative AI and Large Language Models (LLM) on NER and NLP in Classics, emphasizing the challenges for independent research posed by the high costs and limited transparency of LLMs.

@article{temel_using_2025,
	title = {Using {Jupyter} {Notebooks} as digital assessment tools: {An} empirical examination of student teachers’ attitudes and skills towards digital assessment},
	volume = {30},
	url = {https://api.semanticscholar.org/CorpusID:277588522},
	journal = {Education and Information Technologies},
	author = {Temel, Güler Yavuz and Barenthien, Julia and Padubrin, Thore},
	year = {2025},
	pages = {18621 -- 18650},
}

@article{huang_how_2025,
	title = {How {Scientists} {Use} {Jupyter} {Notebooks}: {Goals}, {Quality} {Attributes}, and {Opportunities}},
	url = {https://api.semanticscholar.org/CorpusID:277066276},
	journal = {2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE)},
	author = {Huang, Ruanqianqian and Ravi, Savitha and He, Michael and Tian, Boyu and Lerner, Sorin and Coblenz, Michael},
	year = {2025},
	pages = {1243--1255},
}

@misc{sokefeld_korpusbasierte_2025,
	title = {Korpusbasierte Übungsmaterialien zu {Kookkurrenzen} und {Kollokationen} im {DWDS}-{Wortprofil}},
	url = {https://www.fdr.uni-hamburg.de/record/17926},
	doi = {10.25592/uhhfdm.17926},
	abstract = {Das Ziel der korpusbasierten Übungsmaterialien zu Kollokationen und Kookkurrenzen im DWDS-Wortprofil ist es, das Erlernen von Deutsch als Fremdsprache mit dem Erlernen von Korpuskompetenzen im germanistischen Hochschulunterricht zu verbinden. Die Materialien wurden im Rahmen der Germanistischen Institutspartnerschaft "Korpusdidaktik für formelhafte Sprache" (KoDi-FS) zwischen der Universität Hamburg und der Universität Vilnius entwickelt (Link zur Projektwebsite) und sind im Kontext der Auslandsgermanistik zu verorten.  Bei den Übungsmaterialien handelt es sich zum einen um ein Handout mit Wortschatz- und Grammatikübungen zu Kookkurrenzen und Kollokationen im DWDS-Wortprofil. Zum anderen wird ein "Spickzettel" zur Verfügung gestellt, in dem grundlegende korpuslinguistische und korpusbezogene Begriffe nachgeschlagen werden können. Der Spickzettel existiert in einer kompakteren Version für Studierende sowie in einer ausführlicheren Version für Lehrende, die weitere Hintergrundinformationen gibt. Die Materialien sollen als „Fahrplan“ gelten und können von Lehrenden entweder direkt genutzt, oder nach den eigenen Vorstellungen und Bedürfnissen der Studierenden angepasst werden. Für eine detaillierte Beschreibung der Materialien siehe Sökefeld/Stulen (i. E.).},
	language = {deu},
	urldate = {2026-02-07},
	author = {Sökefeld, Carla and Stulen, Anastasiia},
	month = sep,
	year = {2025},
	keywords = {Auslandsgermanistik, Corpus Literacy, DaF, Kollokation, Kookkurrenz, Korpusdidaktik, Korpuskompetenzen, Korpuslinguistik},
}

Das Ziel der korpusbasierten Übungsmaterialien zu Kollokationen und Kookkurrenzen im DWDS-Wortprofil ist es, das Erlernen von Deutsch als Fremdsprache mit dem Erlernen von Korpuskompetenzen im germanistischen Hochschulunterricht zu verbinden. Die Materialien wurden im Rahmen der Germanistischen Institutspartnerschaft "Korpusdidaktik für formelhafte Sprache" (KoDi-FS) zwischen der Universität Hamburg und der Universität Vilnius entwickelt (Link zur Projektwebsite) und sind im Kontext der Auslandsgermanistik zu verorten.  Bei den Übungsmaterialien handelt es sich zum einen um ein Handout mit Wortschatz- und Grammatikübungen zu Kookkurrenzen und Kollokationen im DWDS-Wortprofil. Zum anderen wird ein "Spickzettel" zur Verfügung gestellt, in dem grundlegende korpuslinguistische und korpusbezogene Begriffe nachgeschlagen werden können. Der Spickzettel existiert in einer kompakteren Version für Studierende sowie in einer ausführlicheren Version für Lehrende, die weitere Hintergrundinformationen gibt. Die Materialien sollen als „Fahrplan“ gelten und können von Lehrenden entweder direkt genutzt, oder nach den eigenen Vorstellungen und Bedürfnissen der Studierenden angepasst werden. Für eine detaillierte Beschreibung der Materialien siehe Sökefeld/Stulen (i. E.).

@misc{zhou_slide2text_2025,
	title = {{Slide2Text}: {Leveraging} {LLMs} for {Personalized} {Textbook} {Generation} from {PowerPoint} {Presentations}},
	shorttitle = {{Slide2Text}},
	url = {http://arxiv.org/abs/2503.17710},
	doi = {10.48550/arXiv.2503.17710},
	abstract = {The rapid advancements in Large Language Models (LLMs) have revolutionized educational technology, enabling innovative approaches to automated and personalized content creation. This paper introduces Slide2Text, a system that leverages LLMs to transform PowerPoint presentations into customized textbooks. By extracting slide content using OCR, organizing it into a coherent structure, and generating tailored materials such as explanations, exercises, and references, Slide2Text streamlines the textbook creation process. Flexible customization options further enhance its adaptability to diverse educational needs. The system highlights the potential of LLMs in modernizing textbook creation and improving educational accessibility. Future developments will explore multimedia inputs and advanced user customization features.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Zhou, Yizhou},
	month = mar,
	year = {2025},
	note = {arXiv:2503.17710 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Electrical Engineering and Systems Science - Image and Video Processing},
}

The rapid advancements in Large Language Models (LLMs) have revolutionized educational technology, enabling innovative approaches to automated and personalized content creation. This paper introduces Slide2Text, a system that leverages LLMs to transform PowerPoint presentations into customized textbooks. By extracting slide content using OCR, organizing it into a coherent structure, and generating tailored materials such as explanations, exercises, and references, Slide2Text streamlines the textbook creation process. Flexible customization options further enhance its adaptability to diverse educational needs. The system highlights the potential of LLMs in modernizing textbook creation and improving educational accessibility. Future developments will explore multimedia inputs and advanced user customization features.

@misc{sun_p2p_2025,
	title = {{P2P}: {Automated} {Paper}-to-{Poster} {Generation} and {Fine}-{Grained} {Benchmark}},
	shorttitle = {{P2P}},
	url = {http://arxiv.org/abs/2505.17104},
	doi = {10.48550/arXiv.2505.17104},
	abstract = {Academic posters are vital for scholarly communication, yet their manual creation is time-consuming. However, automated academic poster generation faces significant challenges in preserving intricate scientific details and achieving effective visual-textual integration. Existing approaches often struggle with semantic richness and structural nuances, and lack standardized benchmarks for evaluating generated academic posters comprehensively. To address these limitations, we introduce P2P, the first flexible, LLM-based multi-agent framework that generates high-quality, HTML-rendered academic posters directly from research papers, demonstrating strong potential for practical applications. P2P employs three specialized agents-for visual element processing, content generation, and final poster assembly-each integrated with dedicated checker modules to enable iterative refinement and ensure output quality. To foster advancements and rigorous evaluation in this domain, we construct and release P2PInstruct, the first large-scale instruction dataset comprising over 30,000 high-quality examples tailored for the academic paper-to-poster generation task. Furthermore, we establish P2PEval, a comprehensive benchmark featuring 121 paper-poster pairs and a dual evaluation methodology (Universal and Fine-Grained) that leverages LLM-as-a-Judge and detailed, human-annotated checklists. Our contributions aim to streamline research dissemination and provide the community with robust tools for developing and evaluating next-generation poster generation systems.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Sun, Tao and Pan, Enhao and Yang, Zhengkai and Sui, Kaixin and Shi, Jiajun and Cheng, Xianfu and Li, Tongliang and Huang, Wenhao and Zhang, Ge and Yang, Jian and Li, Zhoujun},
	month = may,
	year = {2025},
	note = {arXiv:2505.17104 [cs]},
	keywords = {Computer Science - Computation and Language, Computer Science - Multimedia},
}

Academic posters are vital for scholarly communication, yet their manual creation is time-consuming. However, automated academic poster generation faces significant challenges in preserving intricate scientific details and achieving effective visual-textual integration. Existing approaches often struggle with semantic richness and structural nuances, and lack standardized benchmarks for evaluating generated academic posters comprehensively. To address these limitations, we introduce P2P, the first flexible, LLM-based multi-agent framework that generates high-quality, HTML-rendered academic posters directly from research papers, demonstrating strong potential for practical applications. P2P employs three specialized agents-for visual element processing, content generation, and final poster assembly-each integrated with dedicated checker modules to enable iterative refinement and ensure output quality. To foster advancements and rigorous evaluation in this domain, we construct and release P2PInstruct, the first large-scale instruction dataset comprising over 30,000 high-quality examples tailored for the academic paper-to-poster generation task. Furthermore, we establish P2PEval, a comprehensive benchmark featuring 121 paper-poster pairs and a dual evaluation methodology (Universal and Fine-Grained) that leverages LLM-as-a-Judge and detailed, human-annotated checklists. Our contributions aim to streamline research dissemination and provide the community with robust tools for developing and evaluating next-generation poster generation systems.

@misc{gaddipati_aissistant_2025,
	title = {{AIssistant}: {An} {Agentic} {Approach} for {Human}--{AI} {Collaborative} {Scientific} {Work} on {Reviews} and {Perspectives} in {Machine} {Learning}},
	shorttitle = {{AIssistant}},
	url = {http://arxiv.org/abs/2509.12282},
	doi = {10.48550/arXiv.2509.12282},
	abstract = {Advances in AI-assisted research have introduced powerful tools for literature retrieval, hypothesis generation, experimentation, and manuscript preparation. However, systems remain fragmented and lack human-centred workflows. To address these gaps, we introduce AIssistant, an agentic, open-source Human-AI collaborative framework designed to simplify the end-to-end creation of scientific workflows. Since our development is still in an early stage, we present here the first experiments with AIssistant for perspective and review research papers in machine learning. Our system integrates modular tools and agents for literature synthesis, section-wise experimentation, citation management, and automatic LaTeX paper text generation, while maintaining human oversight at every stage to ensure accuracy, coherence, and scholarly rigour. We conducted a comprehensive evaluation across three layers: (1) Independent Human Review, following NeurIPS double-blind standards; (2) Automated LLM Review, using GPT-5 as a scalable human review proxy; and (3) Program Chair Oversight, where the chair monitors the entire review process and makes final validation and acceptance decisions. The results demonstrate that AIssistant improves drafting efficiency and thematic consistency. Nonetheless, Human-AI collaboration remains essential for maintaining factual correctness, methodological soundness, and ethical compliance. Despite its effectiveness, we identify key limitations, including hallucinated citations, difficulty adapting to dynamic paper structures, and incomplete integration of multimodal content.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Gaddipati, Sasi Kiran and Keya, Farhana and Rabby, Gollam and Auer, Sören},
	month = sep,
	year = {2025},
	note = {arXiv:2509.12282 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Machine Learning},
}

Advances in AI-assisted research have introduced powerful tools for literature retrieval, hypothesis generation, experimentation, and manuscript preparation. However, systems remain fragmented and lack human-centred workflows. To address these gaps, we introduce AIssistant, an agentic, open-source Human-AI collaborative framework designed to simplify the end-to-end creation of scientific workflows. Since our development is still in an early stage, we present here the first experiments with AIssistant for perspective and review research papers in machine learning. Our system integrates modular tools and agents for literature synthesis, section-wise experimentation, citation management, and automatic LaTeX paper text generation, while maintaining human oversight at every stage to ensure accuracy, coherence, and scholarly rigour. We conducted a comprehensive evaluation across three layers: (1) Independent Human Review, following NeurIPS double-blind standards; (2) Automated LLM Review, using GPT-5 as a scalable human review proxy; and (3) Program Chair Oversight, where the chair monitors the entire review process and makes final validation and acceptance decisions. The results demonstrate that AIssistant improves drafting efficiency and thematic consistency. Nonetheless, Human-AI collaboration remains essential for maintaining factual correctness, methodological soundness, and ethical compliance. Despite its effectiveness, we identify key limitations, including hallucinated citations, difficulty adapting to dynamic paper structures, and incomplete integration of multimodal content.

@misc{xie_slidebot_2025,
	title = {{SlideBot}: {A} {Multi}-{Agent} {Framework} for {Generating} {Informative}, {Reliable}, {Multi}-{Modal} {Presentations}},
	shorttitle = {{SlideBot}},
	url = {http://arxiv.org/abs/2511.09804},
	doi = {10.48550/arXiv.2511.09804},
	abstract = {Large Language Models (LLMs) have shown immense potential in education, automating tasks like quiz generation and content summarization. However, generating effective presentation slides introduces unique challenges due to the complexity of multimodal content creation and the need for precise, domain-specific information. Existing LLM-based solutions often fail to produce reliable and informative outputs, limiting their educational value. To address these limitations, we introduce SlideBot - a modular, multi-agent slide generation framework that integrates LLMs with retrieval, structured planning, and code generation. SlideBot is organized around three pillars: informativeness, ensuring deep and contextually grounded content; reliability, achieved by incorporating external sources through retrieval; and practicality, which enables customization and iterative feedback through instructor collaboration. It incorporates evidence-based instructional design principles from Cognitive Load Theory (CLT) and the Cognitive Theory of Multimedia Learning (CTML), using structured planning to manage intrinsic load and consistent visual macros to reduce extraneous load and enhance dual-channel learning. Within the system, specialized agents collaboratively retrieve information, summarize content, generate figures, and format slides using LaTeX, aligning outputs with instructor preferences through interactive refinement. Evaluations from domain experts and students in AI and biomedical education show that SlideBot consistently enhances conceptual accuracy, clarity, and instructional value. These findings demonstrate SlideBot's potential to streamline slide preparation while ensuring accuracy, relevance, and adaptability in higher education.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Xie, Eric and Waterfield, Danielle and Kennedy, Michael and Zhang, Aidong},
	month = nov,
	year = {2025},
	note = {arXiv:2511.09804 [cs]},
	keywords = {Computer Science - Artificial Intelligence},
}

Large Language Models (LLMs) have shown immense potential in education, automating tasks like quiz generation and content summarization. However, generating effective presentation slides introduces unique challenges due to the complexity of multimodal content creation and the need for precise, domain-specific information. Existing LLM-based solutions often fail to produce reliable and informative outputs, limiting their educational value. To address these limitations, we introduce SlideBot - a modular, multi-agent slide generation framework that integrates LLMs with retrieval, structured planning, and code generation. SlideBot is organized around three pillars: informativeness, ensuring deep and contextually grounded content; reliability, achieved by incorporating external sources through retrieval; and practicality, which enables customization and iterative feedback through instructor collaboration. It incorporates evidence-based instructional design principles from Cognitive Load Theory (CLT) and the Cognitive Theory of Multimedia Learning (CTML), using structured planning to manage intrinsic load and consistent visual macros to reduce extraneous load and enhance dual-channel learning. Within the system, specialized agents collaboratively retrieve information, summarize content, generate figures, and format slides using LaTeX, aligning outputs with instructor preferences through interactive refinement. Evaluations from domain experts and students in AI and biomedical education show that SlideBot consistently enhances conceptual accuracy, clarity, and instructional value. These findings demonstrate SlideBot's potential to streamline slide preparation while ensuring accuracy, relevance, and adaptability in higher education.

@misc{vu_agenteval_2025,
	title = {{AgentEval}: {Generative} {Agents} as {Reliable} {Proxies} for {Human} {Evaluation} of {AI}-{Generated} {Content}},
	shorttitle = {{AgentEval}},
	url = {http://arxiv.org/abs/2512.08273},
	doi = {10.48550/arXiv.2512.08273},
	abstract = {Modern businesses are increasingly challenged by the time and expense required to generate and assess high-quality content. Human writers face time constraints, and extrinsic evaluations can be costly. While Large Language Models (LLMs) offer potential in content creation, concerns about the quality of AI-generated content persist. Traditional evaluation methods, like human surveys, further add operational costs, highlighting the need for efficient, automated solutions. This research introduces Generative Agents as a means to tackle these challenges. These agents can rapidly and cost-effectively evaluate AI-generated content, simulating human judgment by rating aspects such as coherence, interestingness, clarity, fairness, and relevance. By incorporating these agents, businesses can streamline content generation and ensure consistent, high-quality output while minimizing reliance on costly human evaluations. The study provides critical insights into enhancing LLMs for producing business-aligned, high-quality content, offering significant advancements in automated content generation and evaluation.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Vu, Thanh and Nayak, Richi and Balasubramaniam, Thiru},
	month = dec,
	year = {2025},
	note = {arXiv:2512.08273 [cs]},
	keywords = {Computer Science - Artificial Intelligence},
}

Modern businesses are increasingly challenged by the time and expense required to generate and assess high-quality content. Human writers face time constraints, and extrinsic evaluations can be costly. While Large Language Models (LLMs) offer potential in content creation, concerns about the quality of AI-generated content persist. Traditional evaluation methods, like human surveys, further add operational costs, highlighting the need for efficient, automated solutions. This research introduces Generative Agents as a means to tackle these challenges. These agents can rapidly and cost-effectively evaluate AI-generated content, simulating human judgment by rating aspects such as coherence, interestingness, clarity, fairness, and relevance. By incorporating these agents, businesses can streamline content generation and ensure consistent, high-quality output while minimizing reliance on costly human evaluations. The study provides critical insights into enhancing LLMs for producing business-aligned, high-quality content, offering significant advancements in automated content generation and evaluation.

@misc{schulz_daidalos-projekt_2025,
	title = {Daidalos-{Projekt} {II}. {KI}-{Infrastruktur} für interdisziplinäre {Forschung} an lateinischen und altgriechischen {Texten}},
	url = {https://zenodo.org/records/17964223},
	doi = {10.5281/zenodo.17964223},
	abstract = {Die Präsentation stellt das Projekt Daidalos vor, eine KI-gestützte Forschungsinfrastruktur für die Arbeit mit lateinischen und altgriechischen Texten in den Geisteswissenschaften. Ausgangspunkt ist Daidalos I (2023–2026), das sich auf den Aufbau nachhaltiger NLP-Infrastruktur, die Evaluation und Aufbereitung von Sprachressourcen sowie die Unterstützung von Forschenden durch partizipative Formate konzentriert. Zentrale Ergebnisse sind u. a. ein systematisches Evaluationsframework (SEFLAG), interoperable Korpora und erste methodische Vergleiche zwischen klassischen NLP-Ansätzen und Large Language Models (LLMs).

Mit Daidalos II (geplant 2027–2030) soll diese Arbeit ausgebaut und disziplinär geöffnet werden, etwa für Mediävistik, Theologie, Geschichts- und Sprachwissenschaften. Angesichts des stark wachsenden KI-Einsatzes in den Digital Humanities adressiert das Projekt den Mangel an digitaler Kompetenz, hohe technische Einstiegshürden und die Gefahr unreflektierter KI-Nutzung. Geplant sind mehr Korpora und Methoden, gezielte LLM-Integration, Sommerschulen, „Train-the-Trainer“-Formate sowie die Entwicklung eines tragfähigen Betriebsmodells.

Ein Schwerpunkt liegt auf der Bereitstellung und Standardisierung hochwertiger Textkorpora (z. B. über offene Repositorien, Linked Open Data und etablierte Textservices) sowie auf methodischer Orientierung für digitale Verfahren wie Lemmatisierung, Dependency Parsing, Word Embeddings, Sentimentanalyse und Textklassifikation. LLMs werden kritisch diskutiert: Ihre Potenziale für niedrigschwellige Nutzung stehen Risiken wie Halluzinationen, Intransparenz, ökologische Kosten und Datenschutzfragen gegenüber. Evaluation, Explainable AI und sorgfältige Integration sind daher zentral.

Abschließend betont die Präsentation, dass digitale Forschungskompetenz – verstanden als Kombination aus Digital, Data und AI Literacy – unverzichtbar ist. Daidalos versteht sich als soziale und technische Infrastruktur, die fundierten, reflektierten und nachhaltigen KI-Einsatz in der Klassischen Philologie und darüber hinaus ermöglicht.},
	language = {deu},
	urldate = {2026-01-05},
	author = {Schulz, Konstantin and Beyer, Andrea},
	month = dec,
	year = {2025},
	keywords = {Artificial Intelligence, Artificial intelligence, Classics, Languages and literature, Natural Language Processing, Natural language processing},
}

Die Präsentation stellt das Projekt Daidalos vor, eine KI-gestützte Forschungsinfrastruktur für die Arbeit mit lateinischen und altgriechischen Texten in den Geisteswissenschaften. Ausgangspunkt ist Daidalos I (2023–2026), das sich auf den Aufbau nachhaltiger NLP-Infrastruktur, die Evaluation und Aufbereitung von Sprachressourcen sowie die Unterstützung von Forschenden durch partizipative Formate konzentriert. Zentrale Ergebnisse sind u. a. ein systematisches Evaluationsframework (SEFLAG), interoperable Korpora und erste methodische Vergleiche zwischen klassischen NLP-Ansätzen und Large Language Models (LLMs). Mit Daidalos II (geplant 2027–2030) soll diese Arbeit ausgebaut und disziplinär geöffnet werden, etwa für Mediävistik, Theologie, Geschichts- und Sprachwissenschaften. Angesichts des stark wachsenden KI-Einsatzes in den Digital Humanities adressiert das Projekt den Mangel an digitaler Kompetenz, hohe technische Einstiegshürden und die Gefahr unreflektierter KI-Nutzung. Geplant sind mehr Korpora und Methoden, gezielte LLM-Integration, Sommerschulen, „Train-the-Trainer“-Formate sowie die Entwicklung eines tragfähigen Betriebsmodells. Ein Schwerpunkt liegt auf der Bereitstellung und Standardisierung hochwertiger Textkorpora (z. B. über offene Repositorien, Linked Open Data und etablierte Textservices) sowie auf methodischer Orientierung für digitale Verfahren wie Lemmatisierung, Dependency Parsing, Word Embeddings, Sentimentanalyse und Textklassifikation. LLMs werden kritisch diskutiert: Ihre Potenziale für niedrigschwellige Nutzung stehen Risiken wie Halluzinationen, Intransparenz, ökologische Kosten und Datenschutzfragen gegenüber. Evaluation, Explainable AI und sorgfältige Integration sind daher zentral. Abschließend betont die Präsentation, dass digitale Forschungskompetenz – verstanden als Kombination aus Digital, Data und AI Literacy – unverzichtbar ist. Daidalos versteht sich als soziale und technische Infrastruktur, die fundierten, reflektierten und nachhaltigen KI-Einsatz in der Klassischen Philologie und darüber hinaus ermöglicht.

@phdthesis{gong_intertextai_2025,
	address = {Cambridge (MA)},
	type = {Bachelors {Thesis}},
	title = {Intertext.{AI}: {Augmented} {Close} {Reading} for {Classical} {Latin} using {AI} for {Intertextual} {Exploration}},
	shorttitle = {Intertext.{AI}},
	url = {https://dash.harvard.edu/handle/1/42719198},
	abstract = {Technological tools for computational language analysis, including applications for ancient languages like classical Latin, continue to improve as artificial intelligence (AI) models become more advanced. However, the subjective process of literary analysis is still primarily manual, relying on commentaries and secondary scholarship. Identifying and interpreting meaningful textual connections, or intertextuality, is labor-intensive and can often present a difficult learning curve for less experienced classicists. Although existing digital platforms offer complex searches by various linguistic qualities such as syntax and phrase similarity, they often do not enable comparisons of the results in their broader contexts, which can provide deeper insights that short excerpts do not reveal. Thus, this thesis proposes the use of AI and visualizations that prioritize contextualization to facilitate the discovery of intertextual correspondences using complex quantitative representations of texts. We introduce a novel web interface, Intertext.AI, that integrates Latin BERT (Bamman and Burns 2020), a machine learning language model trained on classical Latin texts, into contextually rich text visualizations to assist classicists in searching for potential intertextual connections. To evaluate the system, we tested its ability to find allusions attested in classical scholarship, investigated a case study on how a reader may use the interface to enhance their close reading, and conducted a user study with 19 participants who explored potential correspondence between two pairs of Latin poems. Intertext.AI identified over 80\% of attested connections from excerpts of Lucan’s Pharsalia, demonstrating the system’s technical efficacy at detecting allusions. Further, while participants did not identify significantly different types or quantities of connections when using Intertext.AI or other tools, they overall found it easier to find and justify potential intertextuality with Intertext.AI, reported higher confidence in their observations identified through the interface, and preferred having access to it during the search process. These findings thus suggest that Intertext.AI facilitates meaningful intertextual discovery and interpretation by fostering literary comparison.},
	language = {en},
	urldate = {2025-12-30},
	school = {Harvard University Engineering and Applied Sciences},
	author = {Gong, Ashley},
	month = jun,
	year = {2025},
}

Technological tools for computational language analysis, including applications for ancient languages like classical Latin, continue to improve as artificial intelligence (AI) models become more advanced. However, the subjective process of literary analysis is still primarily manual, relying on commentaries and secondary scholarship. Identifying and interpreting meaningful textual connections, or intertextuality, is labor-intensive and can often present a difficult learning curve for less experienced classicists. Although existing digital platforms offer complex searches by various linguistic qualities such as syntax and phrase similarity, they often do not enable comparisons of the results in their broader contexts, which can provide deeper insights that short excerpts do not reveal. Thus, this thesis proposes the use of AI and visualizations that prioritize contextualization to facilitate the discovery of intertextual correspondences using complex quantitative representations of texts. We introduce a novel web interface, Intertext.AI, that integrates Latin BERT (Bamman and Burns 2020), a machine learning language model trained on classical Latin texts, into contextually rich text visualizations to assist classicists in searching for potential intertextual connections. To evaluate the system, we tested its ability to find allusions attested in classical scholarship, investigated a case study on how a reader may use the interface to enhance their close reading, and conducted a user study with 19 participants who explored potential correspondence between two pairs of Latin poems. Intertext.AI identified over 80% of attested connections from excerpts of Lucan’s Pharsalia, demonstrating the system’s technical efficacy at detecting allusions. Further, while participants did not identify significantly different types or quantities of connections when using Intertext.AI or other tools, they overall found it easier to find and justify potential intertextuality with Intertext.AI, reported higher confidence in their observations identified through the interface, and preferred having access to it during the search process. These findings thus suggest that Intertext.AI facilitates meaningful intertextual discovery and interpretation by fostering literary comparison.

@inproceedings{gong_augmented_2025,
	address = {Stroudsburg, PA},
	title = {Augmented {Close} {Reading} for {Classical} {Latin} using {BERT} for {Intertextual} {Exploration}},
	url = {https://aclanthology.org/2025.nlp4dh-1.35.pdf},
	abstract = {Intertextuality, the connection between texts, is a critical literary concept for analyzing classical Latin works. Given the emergence of AI in digital humanities, this paper presents Intertext.AI, a novel interface that leverages Latin BERT (Bamman and Burns, 2020), a BERT model trained on classical Latin texts, and contextually rich visualizations to help classicists find potential intertextual connections. Intertext.AI identified over 80\% of attested allusions from excerpts of Lucan’s Pharsalia, demonstrating the system’s technical efficacy. Our findings from a user study with 19 participants also suggest that Intertext.AI fosters intertextual discovery and interpretation more easily than other tools. While participants did not identify significantly different types or quantities of connections when using Intertext.AI or other tools, they overall found finding and justifying potential intertextuality easier with Intertext.AI, reported higher confidence in their observations from Intertext.AI, and preferred having access to it during the search process.},
	language = {en},
	booktitle = {Proceedings of the 5th {International} {Conference} on {Natural} {Language} {Processing} for {Digital} {Humanities}},
	publisher = {Association for Computational Linguistics},
	author = {Gong, Ashley and Gero, Katy Ilonka and Schiefsky, Mark},
	year = {2025},
	pages = {403--417},
}

Intertextuality, the connection between texts, is a critical literary concept for analyzing classical Latin works. Given the emergence of AI in digital humanities, this paper presents Intertext.AI, a novel interface that leverages Latin BERT (Bamman and Burns, 2020), a BERT model trained on classical Latin texts, and contextually rich visualizations to help classicists find potential intertextual connections. Intertext.AI identified over 80% of attested allusions from excerpts of Lucan’s Pharsalia, demonstrating the system’s technical efficacy. Our findings from a user study with 19 participants also suggest that Intertext.AI fosters intertextual discovery and interpretation more easily than other tools. While participants did not identify significantly different types or quantities of connections when using Intertext.AI or other tools, they overall found finding and justifying potential intertextuality easier with Intertext.AI, reported higher confidence in their observations from Intertext.AI, and preferred having access to it during the search process.

@article{singh_human-computer_2025,
	title = {Human-{Computer} {Interaction}: {A} {Review} of {Usability}, {Design}, and {Accessibility} {Trends}},
	volume = {1},
	copyright = {Copyright (c) 2025 Ankur  singh (Author)},
	issn = {3079-5761},
	shorttitle = {Human-{Computer} {Interaction}},
	url = {https://globaltrendsst.com/index.php/GTST/article/view/34},
	doi = {10.70445/gtst.1.4.2025.25-46},
	abstract = {A critical review of the Human-Computer Interaction (HCI), and as such prime areas of HCI: usability, interface design, accessibility, cognitive and domain specific uses of HCI as well as evaluation techniques of HCI and ethical considerations. With the growing exposures of all interactive technologies to daily life, systems that can be both access able inclusiveness, and ethically contemplated are more than ever in demand. This article has highlighted the importance of user-centered design in enhancing performance and satisfaction of users when looking at the issues multimodal interaction, cultural sensitivity and the digital divide. Design assessment tools like usability testing, eye-tracking are addressed as tools that are needed to assess and assist in improving design. Further, the paper identifies augmented accountability on the part of HCI practitioners in matters of privacy, equity, and access. This review offers an in depth perspective of how HCI can be leveraged to establish a more productive, fair, and more human technological future with an eventual zing review of these axes of borrowing coming near.},
	language = {en},
	number = {4},
	urldate = {2025-12-30},
	journal = {Global Trends in Science and Technology},
	author = {Singh, Ankur},
	month = aug,
	year = {2025},
	keywords = {Accessibility, Congnitive Factors, HCI, Interface Design, Usability, User Experience},
	pages = {25--46},
}

A critical review of the Human-Computer Interaction (HCI), and as such prime areas of HCI: usability, interface design, accessibility, cognitive and domain specific uses of HCI as well as evaluation techniques of HCI and ethical considerations. With the growing exposures of all interactive technologies to daily life, systems that can be both access able inclusiveness, and ethically contemplated are more than ever in demand. This article has highlighted the importance of user-centered design in enhancing performance and satisfaction of users when looking at the issues multimodal interaction, cultural sensitivity and the digital divide. Design assessment tools like usability testing, eye-tracking are addressed as tools that are needed to assess and assist in improving design. Further, the paper identifies augmented accountability on the part of HCI practitioners in matters of privacy, equity, and access. This review offers an in depth perspective of how HCI can be leveraged to establish a more productive, fair, and more human technological future with an eventual zing review of these axes of borrowing coming near.

@inproceedings{assis_integrating_2025,
	title = {Integrating {Participatory} {Design} and {Dual}-track {Software} {Development} {Process}: {A} {Case} {Study} {From} an {Intelligent} {Mathematics} {Tutoring} {System}},
	copyright = {Copyright (c)},
	issn = {0000-0000},
	shorttitle = {Integrating {Participatory} {Design} and {Dual}-track {Software} {Development} {Process}},
	url = {https://sol.sbc.org.br/index.php/cibse/article/view/35290},
	doi = {10.5753/cibse.2025.35290},
	abstract = {The demand for educational technology has increased research into Intelligent Tutoring Systems (ITS) for low-connectivity regions. However, misalignment between requirement discovery and software development hinders adoption. This case study examines the development of an unplugged ITS using dual-track agile development and participatory design to ensure continuous alignment with pedagogical needs. This study contributes to software engineering by demonstrating how we integrated dual-track development and participatory design to enhance agility while maintaining a user-centered approach in a project with a scalable model for ITS applications in low-resource environments.},
	language = {en},
	urldate = {2025-12-13},
	booktitle = {Congresso {Ibero}-{Americano} em {Engenharia} de {Software} ({CIbSE})},
	publisher = {SBC},
	author = {Assis, João Victor and Guerino, Guilherme and Rodrigues, Luiz and Macario, Valmir and Marinho, Marcelo},
	month = may,
	year = {2025},
	pages = {30--44},
}

The demand for educational technology has increased research into Intelligent Tutoring Systems (ITS) for low-connectivity regions. However, misalignment between requirement discovery and software development hinders adoption. This case study examines the development of an unplugged ITS using dual-track agile development and participatory design to ensure continuous alignment with pedagogical needs. This study contributes to software engineering by demonstrating how we integrated dual-track development and participatory design to enhance agility while maintaining a user-centered approach in a project with a scalable model for ITS applications in low-resource environments.

@inproceedings{ma_cocreation_2025,
	address = {Washington D.C.},
	title = {Co‐{Creation} in {Context}: {Participatory} {Approaches} to {Digital} {Humanities} and {Cultural} {Heritage} {Work}},
	shorttitle = {Co‐{Creation} in {Context}},
	url = {https://asistdl.onlinelibrary.wiley.com/doi/10.1002/pra2.1379},
	doi = {10.1002/pra2.1379},
	language = {en},
	urldate = {2025-12-13},
	booktitle = {Proceedings of the {Association} for {Information} {Science} and {Technology}},
	author = {Ma, Rongqian and Chen, Annie T. and Bossaller, Jenny and Boyles, Christina and Donaldson, Devan Ray},
	year = {2025},
	pages = {1264--1269},
}

@misc{galka_annotating_2025,
	title = {Annotating, {Projecting}, and {Interpreting} {Named} {Entities} in {Digital} {Scholarly} {Editions} with {LLMs}},
	url = {https://www.researchsquare.com/article/rs-7175875/v1},
	doi = {10.21203/rs.3.rs-7175875/v1},
	abstract = {This paper explores the use of large language models (LLMs) to enhance semantic annotation and annotation projection in digital scholarly editions (DSEs), focusing on historical ego-documents. Using the TEI/XML-encoded French memoirs of Countess Luise Charlotte of Schwerin (1684\–1732) and their German translation as a case study, we evaluate LLMs for Named Entity Recognition (NER), annotation transfer across aligned bilingual texts, and the extraction of interpersonal relationships. A comparative analysis with a traditional NER framework shows that LLMs significantly outperform baseline models, particularly in recognizing complex person references, such as non-rigid designators and nested entities. For annotation projection, we demonstrate that LLMs can reliably transfer entity annotations between French and German texts without intermediate alignment layers, achieving over 97\% of correct projected entities using zero-shot prompting. Additionally, a pilot experiment illustrates the potential of LLMs for structured relationship modeling. The analysis of the errors puts further emphasis on the question of our intentions as editors when translating and indexing texts.},
	urldate = {2025-12-28},
	publisher = {Research Square},
	author = {Galka, Selina and Vogeler, Georg},
	month = aug,
	year = {2025},
	note = {ISSN: 2693-5015},
}

This paper explores the use of large language models (LLMs) to enhance semantic annotation and annotation projection in digital scholarly editions (DSEs), focusing on historical ego-documents. Using the TEI/XML-encoded French memoirs of Countess Luise Charlotte of Schwerin (1684–1732) and their German translation as a case study, we evaluate LLMs for Named Entity Recognition (NER), annotation transfer across aligned bilingual texts, and the extraction of interpersonal relationships. A comparative analysis with a traditional NER framework shows that LLMs significantly outperform baseline models, particularly in recognizing complex person references, such as non-rigid designators and nested entities. For annotation projection, we demonstrate that LLMs can reliably transfer entity annotations between French and German texts without intermediate alignment layers, achieving over 97% of correct projected entities using zero-shot prompting. Additionally, a pilot experiment illustrates the potential of LLMs for structured relationship modeling. The analysis of the errors puts further emphasis on the question of our intentions as editors when translating and indexing texts.

@misc{trilcke_agentic_2025,
	title = {Agentic {DraCor} and the {Art} of {Docstring} {Engineering}: {Evaluating} {MCP}-empowered {LLM} {Usage} of the {DraCor} {API}},
	shorttitle = {Agentic {DraCor} and the {Art} of {Docstring} {Engineering}},
	url = {http://arxiv.org/abs/2508.13774},
	doi = {10.48550/arXiv.2508.13774},
	abstract = {This paper reports on the implementation and evaluation of a Model Context Protocol (MCP) server for DraCor, enabling Large Language Models (LLM) to autonomously interact with the DraCor API. We conducted experiments focusing on tool selection and application by the LLM, employing a qualitative approach that includes systematic observation of prompts to understand how LLMs behave when using MCP tools, evaluating "Tool Correctness", "Tool-Calling Efficiency", and "Tool-Use Reliability". Our findings highlight the importance of "Docstring Engineering", defined as reflexively crafting tool documentation to optimize LLM-tool interaction. Our experiments demonstrate both the promise of agentic AI for research in Computational Literary Studies and the essential infrastructure development needs for reliable Digital Humanities infrastructures.},
	urldate = {2025-12-28},
	publisher = {arXiv},
	author = {Trilcke, Peer and Börner, Ingo and Sluyter-Gäthje, Henny and Skorinkin, Daniil and Fischer, Frank and Milling, Carsten},
	month = aug,
	year = {2025},
	note = {arXiv:2508.13774 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Software Engineering},
}

This paper reports on the implementation and evaluation of a Model Context Protocol (MCP) server for DraCor, enabling Large Language Models (LLM) to autonomously interact with the DraCor API. We conducted experiments focusing on tool selection and application by the LLM, employing a qualitative approach that includes systematic observation of prompts to understand how LLMs behave when using MCP tools, evaluating "Tool Correctness", "Tool-Calling Efficiency", and "Tool-Use Reliability". Our findings highlight the importance of "Docstring Engineering", defined as reflexively crafting tool documentation to optimize LLM-tool interaction. Our experiments demonstrate both the promise of agentic AI for research in Computational Literary Studies and the essential infrastructure development needs for reliable Digital Humanities infrastructures.

@misc{korbak_chain_2025,
	title = {Chain of thought monitorability: {A} new and fragile opportunity for {AI} safety},
	url = {https://arxiv.org/pdf/2507.11473},
	publisher = {arXiv},
	author = {Korbak, Tomek and Balesni, Mikita and Barnes, Elizabeth and Bengio, Yoshua and Benton, Joe and Bloom, Joseph and Chen, Mark and Cooney, Alan and Dafoe, Allan and Dragan, Anca},
	year = {2025},
}

@misc{cobb_module_2025,
	type = {Open {Educational} {Resource}},
	title = {Module 1: {Introduction} to {Digital} {Classics} / {Einführung} in die {Digital} {Classics}},
	shorttitle = {Module 1},
	url = {https://zenodo.org/records/17724155},
	doi = {10.5281/zenodo.17724154},
	abstract = {Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/). 

Die OER-Ressource wird in zwei Fssungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszei beträgt knapp 2h. Für weitere Informationen s. readme.txt.

Inhalt: 



Was ist Digital Classics?

Digitale Ressourcen (Texte, Übersetzungen) nutzen

Texte alignieren und übersetzen

Texte skandieren

Textüberlieferung


----

The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/).

The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 2 hours to complete. For further information, see readme.txt.

Content:



What is Digital Classics?

Using Digital Texts \& Translation Resources

Text Alignment \& Translation

Scansion

Transmission},
	language = {en},
	urldate = {2025-11-27},
	publisher = {Zenodo},
	author = {Cobb, George},
	collaborator = {Beyer, Andrea},
	month = nov,
	year = {2025},
	doi = {10.5281/zenodo.17724155},
	keywords = {Digital Classics, Klassische Philologie, Latin Philology},
}

Der Moodle-basierte Lernbaustein ist Teil des Projektes "Studienbegleitende Grundbildung Digital Literacy im BA Lateinische Philologie", das von der Medienkommission der Humboldt-Universität zu Berlin gefördert wird (2024-2026). Das Projekt ist assoziiert mit dem DFG-Projekt Daidalos (https://daidalos-projekt.de/).  Die OER-Ressource wird in zwei Fssungen (en, de) angeboten und kann in jede aktuelle Moodle-Plattform importiert werden. Der Lernbaustein ist als Selbstlernkurs angelegt. Seine Bearbeitungszei beträgt knapp 2h. Für weitere Informationen s. readme.txt. Inhalt:  Was ist Digital Classics? Digitale Ressourcen (Texte, Übersetzungen) nutzen Texte alignieren und übersetzen Texte skandieren Textüberlieferung —- The Moodle-based learning module is part of the project ‘Basic Digital Literacy Training in the BA Latin Philology Programme’, which is funded by the Media Commission of Humboldt University in Berlin (2024-2026). The project is associated with the DFG project Daidalos (https://daidalos-projekt.de/). The OER resource is available in two versions (en, ger) and can be imported into any current Moodle platform. The learning module is designed as a self-study course. It takes just under 2 hours to complete. For further information, see readme.txt. Content: What is Digital Classics? Using Digital Texts & Translation Resources Text Alignment & Translation Scansion Transmission

@misc{beyer_neu_2025,
	address = {Berlin},
	type = {Zenodo},
	title = {Neu ist immer besser? - {Texteditionen} und digitale {Methoden}},
	url = {https://zenodo.org/records/14771606},
	doi = {10.5281/zenodo.14771606},
	abstract = {Der Impulsvortrag untersucht die Relevanz aktueller Texteditionen für die digitale Sprachverarbeitung anhand der fiktiven Rede Invectiva in M. Tullium Ciceronem aus dem frühen 1. Jahrhundert n. Chr. Er vergleicht zwei leicht voneinander abweichende Editionen – die von Schöne \& Eisenhut (1969) und die von Novokhatko (2009) – hinsichtlich ihres Einflusses auf die Ergebnisse von Named Entity Recognition (NER) und Sentiment-Analyse. Die Analyse zeigt, dass sowohl bei der Entitätsextraktion als auch bei der Sentiment-Bewertung keine signifikanten Unterschiede zwischen den Editionen auftreten, selbst bei unterschiedlichen Lemmatisierungen. Die Ergebnisse deuten darauf hin, dass ältere, öffentlich zugängliche Editionen für digitale Forschungsmethoden ausreichend sind. Die Studie betont, dass digitale Methoden traditionelle philologische Forschung durch empirische, systematische Analysen ergänzen können. Fazit: Forschungsfragen lassen sich auch mit älteren Editionen zuverlässig computerlinguistisch bearbeiten.},
	language = {Deutsch},
	author = {Beyer, Andrea and Kotschka, Florian},
	month = jan,
	year = {2025},
	keywords = {bold, other},
}

Der Impulsvortrag untersucht die Relevanz aktueller Texteditionen für die digitale Sprachverarbeitung anhand der fiktiven Rede Invectiva in M. Tullium Ciceronem aus dem frühen 1. Jahrhundert n. Chr. Er vergleicht zwei leicht voneinander abweichende Editionen – die von Schöne & Eisenhut (1969) und die von Novokhatko (2009) – hinsichtlich ihres Einflusses auf die Ergebnisse von Named Entity Recognition (NER) und Sentiment-Analyse. Die Analyse zeigt, dass sowohl bei der Entitätsextraktion als auch bei der Sentiment-Bewertung keine signifikanten Unterschiede zwischen den Editionen auftreten, selbst bei unterschiedlichen Lemmatisierungen. Die Ergebnisse deuten darauf hin, dass ältere, öffentlich zugängliche Editionen für digitale Forschungsmethoden ausreichend sind. Die Studie betont, dass digitale Methoden traditionelle philologische Forschung durch empirische, systematische Analysen ergänzen können. Fazit: Forschungsfragen lassen sich auch mit älteren Editionen zuverlässig computerlinguistisch bearbeiten.

@misc{beyer_llm_2025,
	address = {Würzburg},
	type = {Zenodo},
	title = {{LLM} \& {NLP} in der {Klassischen} {Philologie}},
	url = {https://zenodo.org/records/14679224},
	doi = {10.5281/zenodo.14679223},
	abstract = {Der Vortrag präsentiert einen Vergleich zwischen dem Einsatz von Natural Language Processing (NLP) und Large Language Models (LLM) in der klassischen Philologie, exemplarisch am Werk des römischen Historikers Sallust. Die Fallstudie untersucht die emotionale und subjektive Färbung in den Monographien Bellum Catilinae und Bellum Iugurthinum hinsichtlich der Charakterisierung der beiden Protagonisten Catilina und Iugurtha. Mittels einer Gegenüberstellung aus klassischer NLP-Methodik – Sentiment-Analyse mit dem LatinAffectus-Lexikon – und der Nutzung verschiedener LLMs (GPT-4o, Gemini-1.5-Pro, Claude-3.5-Sonnet, Qwen, Grok-beta, Llama-3.1) wird die Darstellung der beiden hostes verglichen. Die Ergebnisse zeigen, dass die LLM-basierten Analysen eine hohe Variabilität in den Ergebnissen zeitigen. Die Beispielstudie unterstreicht die Notwendigkeit einer methodischen Reflexion, der Entwicklung spezifischer Benchmarks für alte Sprachen und die Integration von Technologien wie RAG (Retrieval Augmented Generation) und LoRA (Low-Rank Adaptation) zur Verbesserung der Datenqualität und der wissenschaftlichen Aussagekraft. Abschließend wird die Bedeutung interdisziplinärer Zusammenarbeit, der Datenkuratierung und der ethischen und strukturellen Implikationen von KI in der Forschung betont.},
	language = {de},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jan,
	year = {2025},
	keywords = {bold, other},
}

Der Vortrag präsentiert einen Vergleich zwischen dem Einsatz von Natural Language Processing (NLP) und Large Language Models (LLM) in der klassischen Philologie, exemplarisch am Werk des römischen Historikers Sallust. Die Fallstudie untersucht die emotionale und subjektive Färbung in den Monographien Bellum Catilinae und Bellum Iugurthinum hinsichtlich der Charakterisierung der beiden Protagonisten Catilina und Iugurtha. Mittels einer Gegenüberstellung aus klassischer NLP-Methodik – Sentiment-Analyse mit dem LatinAffectus-Lexikon – und der Nutzung verschiedener LLMs (GPT-4o, Gemini-1.5-Pro, Claude-3.5-Sonnet, Qwen, Grok-beta, Llama-3.1) wird die Darstellung der beiden hostes verglichen. Die Ergebnisse zeigen, dass die LLM-basierten Analysen eine hohe Variabilität in den Ergebnissen zeitigen. Die Beispielstudie unterstreicht die Notwendigkeit einer methodischen Reflexion, der Entwicklung spezifischer Benchmarks für alte Sprachen und die Integration von Technologien wie RAG (Retrieval Augmented Generation) und LoRA (Low-Rank Adaptation) zur Verbesserung der Datenqualität und der wissenschaftlichen Aussagekraft. Abschließend wird die Bedeutung interdisziplinärer Zusammenarbeit, der Datenkuratierung und der ethischen und strukturellen Implikationen von KI in der Forschung betont.

@misc{beyer_digital_2025,
	address = {Universität Freiburg},
	title = {Digital {Classics} und {KI}},
	url = {https://zenodo.org/records/15721794},
	abstract = {Der Vortrag stellt den Einsatz künstlicher Intelligenz (KI) und digitaler Methoden in den Digital Classics vor. Er beschreibt die zentralen Schritte der Digitalisierung antiker Texte, beginnend mit der Erfassung von Handschriften mittels Scans oder Fotos, gefolgt von der automatischen Verarbeitung durch Optical Character Recognition (OCR) oder Handwritten Text Recognition (HTR). Die anschließende Normalisierung des Transkripts, einschließlich der Auflösung von Ligaturen, der Vereinheitlichung von Schreibweisen und der Wiederherstellung von Worttrennungen, bildet die Grundlage für die maschinelle Sprachverarbeitung. Nach der Bereinigung des Textes – einschließlich Segmentierung, Tokenisierung und Lemmatisierung – werden verschiedene Analysemethoden des Natural Language Processing (NLP) eingesetzt, darunter Part-of-Speech-Tagging, Word Embeddings, Text Classification, Sentiment Analysis und Topic Modeling. Anhand konkreter Fallstudien wird gezeigt, wie diese Methoden zur Autorschaftsbestimmung, zur Analyse von Wortfeldern, zur Erkennung von Intertextualität zwischen lateinischen und griechischen Texten sowie zur Themenanalyse in Werken wie der Aeneis eingesetzt werden können. Die Ergebnisse verdeutlichen das Potential von KI-Modellen zur Unterstützung und Erweiterung klassisch-philologischer Forschung, wobei auch Grenzen und Herausforderungen bei der Interpretation der Ergebnisse diskutiert werden.},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jun,
	year = {2025},
}

Der Vortrag stellt den Einsatz künstlicher Intelligenz (KI) und digitaler Methoden in den Digital Classics vor. Er beschreibt die zentralen Schritte der Digitalisierung antiker Texte, beginnend mit der Erfassung von Handschriften mittels Scans oder Fotos, gefolgt von der automatischen Verarbeitung durch Optical Character Recognition (OCR) oder Handwritten Text Recognition (HTR). Die anschließende Normalisierung des Transkripts, einschließlich der Auflösung von Ligaturen, der Vereinheitlichung von Schreibweisen und der Wiederherstellung von Worttrennungen, bildet die Grundlage für die maschinelle Sprachverarbeitung. Nach der Bereinigung des Textes – einschließlich Segmentierung, Tokenisierung und Lemmatisierung – werden verschiedene Analysemethoden des Natural Language Processing (NLP) eingesetzt, darunter Part-of-Speech-Tagging, Word Embeddings, Text Classification, Sentiment Analysis und Topic Modeling. Anhand konkreter Fallstudien wird gezeigt, wie diese Methoden zur Autorschaftsbestimmung, zur Analyse von Wortfeldern, zur Erkennung von Intertextualität zwischen lateinischen und griechischen Texten sowie zur Themenanalyse in Werken wie der Aeneis eingesetzt werden können. Die Ergebnisse verdeutlichen das Potential von KI-Modellen zur Unterstützung und Erweiterung klassisch-philologischer Forschung, wobei auch Grenzen und Herausforderungen bei der Interpretation der Ergebnisse diskutiert werden.

@misc{beyer_ner_2025,
	address = {Universität Freiburg},
	type = {Zenodo},
	title = {{NER}: {Von} lexikon- bis chatbasiert},
	url = {https://zenodo.org/records/15722725},
	doi = {10.5281/zenodo.15722724},
	abstract = {Named Entity Recognition (NER) stellt eine zentrale Methode des Information Retrieval dar, die darauf abzielt, in Textkorpora eindeutig identifizierbare Entitäten wie Personen, Orte, Organisationen und politische oder religiöse Gruppen zu erkennen und zu klassifizieren. Der Vortrag untersucht die Entwicklung und Anwendung von NER-Methoden im Kontext der historischen Sprachen Latein und Altgriechisch und vergleicht traditionelle Ansätze – wie lexikonbasierte, regelbasierte und merkmalbasierte Verfahren – mit modernen neuronalen Netzwerken und KI-gestützten Chatbots. Am Beispiel des pseudo-sallustischen Werks In Ciceronem wird der Einsatz von NER zur Analyse von Diskursstruktur, Adressatensystemen und sozialen Netzwerken vorgestellt. Die Studie zeigt, dass Chatbots wie Claude-Sonnet-4 eine hohe Genauigkeit bei der Identifikation und Klassifikation von Entitäten, jedoch bei der Aggregation und Interpretation von Daten sowie bei der Visualisierung von Ergebnissen erhebliche Einschränkungen aufweisen. Die Analyse verdeutlicht zudem die Herausforderungen durch sprachliche Variabilität, Ambiguität und fehlende Ressourcen in Low-Resource-Languages. Zuletzt werden potentielle Forschungsanwendungen in der digitalen Edition, Quellenforschung und computergestützten literarischen Studien aufgezeigt, wobei die Notwendigkeit standardisierter Annotationen und erweiterter Korpora betont wird.},
	language = {de},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jun,
	year = {2025},
}

Named Entity Recognition (NER) stellt eine zentrale Methode des Information Retrieval dar, die darauf abzielt, in Textkorpora eindeutig identifizierbare Entitäten wie Personen, Orte, Organisationen und politische oder religiöse Gruppen zu erkennen und zu klassifizieren. Der Vortrag untersucht die Entwicklung und Anwendung von NER-Methoden im Kontext der historischen Sprachen Latein und Altgriechisch und vergleicht traditionelle Ansätze – wie lexikonbasierte, regelbasierte und merkmalbasierte Verfahren – mit modernen neuronalen Netzwerken und KI-gestützten Chatbots. Am Beispiel des pseudo-sallustischen Werks In Ciceronem wird der Einsatz von NER zur Analyse von Diskursstruktur, Adressatensystemen und sozialen Netzwerken vorgestellt. Die Studie zeigt, dass Chatbots wie Claude-Sonnet-4 eine hohe Genauigkeit bei der Identifikation und Klassifikation von Entitäten, jedoch bei der Aggregation und Interpretation von Daten sowie bei der Visualisierung von Ergebnissen erhebliche Einschränkungen aufweisen. Die Analyse verdeutlicht zudem die Herausforderungen durch sprachliche Variabilität, Ambiguität und fehlende Ressourcen in Low-Resource-Languages. Zuletzt werden potentielle Forschungsanwendungen in der digitalen Edition, Quellenforschung und computergestützten literarischen Studien aufgezeigt, wobei die Notwendigkeit standardisierter Annotationen und erweiterter Korpora betont wird.

@inproceedings{neuroth_quadriga-datenkompetenzframework_2025,
	address = {Chemnitz},
	title = {Das {QUADRIGA}-{Datenkompetenzframework} als {Basis} für die {Entwicklung} von {Lehr}- und {Lernressourcen}},
	url = {https://zenodo.org/records/14925620},
	doi = {https://zenodo.org/records/14925620},
	urldate = {2025-07-06},
	booktitle = {Datenströme und {Kulturoasen} – {Die} {Informationswissenschaft} als {Bindeglied} zwischen den {Informationswelten}},
	author = {Neuroth, Heike and Petras, Vivien and Schnaitter, Hannes and Seltmann, Melanie and Walter, Paul},
	month = mar,
	year = {2025},
	pages = {316--331},
}

@misc{petras_quadriga_2025,
	title = {{QUADRIGA} {Datenkompetenzframework}},
	url = {https://zenodo.org/records/15058057},
	language = {deu},
	urldate = {2025-07-06},
	author = {Petras, Vivien and Neuroth, Heike and Seltmann, Melanie and Schnaitter, Hannes and Walter, Paul},
	month = mar,
	year = {2025},
}

@misc{european_commission_empowering_2025,
	title = {Empowering {Learners} for the {Age} of {AI}. {An} {AI} {Literacy} {Framework} for {Primary} and {Secondary} {Education}},
	url = {https://ailiteracyframework.org/wp-content/uploads/2025/05/AILitFramework_ReviewDraft.pdf},
	editor = {European Commission and OECD},
	year = {2025},
}

@misc{dreyer_kunstliche_2025,
	title = {Künstliche {Intelligenz} in der {Hochschullandschaft}: {Balance} zwischen {Innovation} und {Selbstbestimmung}},
	shorttitle = {Künstliche {Intelligenz} in der {Hochschullandschaft}},
	url = {https://hochschulforumdigitalisierung.de/ki-hochschule/},
	abstract = {In diesem Blogbeitrag berichtet Malte Dreyer, Direktor des Computer- und Medienservice der Humboldt-Universität zu Berlin, über den aktuellen Stand der Anwendung von KI-Programmen an Hochschulen und Universitäten.},
	language = {de-DE},
	urldate = {2025-11-17},
	journal = {Hochschulforum Digitalisierung},
	author = {Dreyer, Malte},
	month = aug,
	year = {2025},
	note = {url: https://hochschulforumdigitalisierung.de/ki-hochschule/},
}

In diesem Blogbeitrag berichtet Malte Dreyer, Direktor des Computer- und Medienservice der Humboldt-Universität zu Berlin, über den aktuellen Stand der Anwendung von KI-Programmen an Hochschulen und Universitäten.

@misc{filograsso_heritrace_2025,
	title = {{HERITRACE} in action: the {ParaText} project as a case study for semantic data management in {Classical} {Philology}},
	shorttitle = {{HERITRACE} in action},
	url = {http://arxiv.org/abs/2508.15556},
	abstract = {HERITRACE is a semantic data editor designed for cultural heritage institutions, addressing the gap between complex Semantic Web technologies and domain expert needs. ParaText Bibliographical Database, a specialized bibliographical database for ancient Greek exegesis, demonstrates HERITRACE's capabilities in Classical Philology. This paper examines how HERITRACE enables non-technical scholars to manage complex semantic data through SHACL-based form generation and validation, while ensuring comprehensive provenance tracking and change management via an OpenCitations Data Model adaptation.},
	urldate = {2025-12-02},
	publisher = {arXiv},
	author = {Filograsso, Francesca and Massari, Arcangelo and Neri, Camillo and Peroni, Silvio},
	month = aug,
	year = {2025},
	note = {arXiv:2508.15556 [cs]},
	keywords = {Computer Science - Digital Libraries},
}

HERITRACE is a semantic data editor designed for cultural heritage institutions, addressing the gap between complex Semantic Web technologies and domain expert needs. ParaText Bibliographical Database, a specialized bibliographical database for ancient Greek exegesis, demonstrates HERITRACE's capabilities in Classical Philology. This paper examines how HERITRACE enables non-technical scholars to manage complex semantic data through SHACL-based form generation and validation, while ensuring comprehensive provenance tracking and change management via an OpenCitations Data Model adaptation.

@misc{beyer_auswertung_2025,
	type = {Zenodo},
	title = {Auswertung der {Evaluationen} im {Projekt}},
	url = {https://zenodo.org/records/17099269},
	doi = {10.5281/zenodo.17099269},
	abstract = {Die Präsentation beschreibt die Ergebnisse verschiedener Evaluationen im Kontext des DFG-Projekts Daidalos. Dazu zählen Umfragen, Interviews und Echtzeit-Überwachung von Interaktionen mit der Projekt-Software. Mit diesen Instrumenten wurden sowohl die vom Projekt durchgeführten Workshops als auch die entwickelte Forschungsinfrastruktur (Web-Anwendung) bewertet.},
	language = {deu},
	urldate = {2025-09-11},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2025},
	keywords = {Artificial Intelligence, Artificial intelligence, Classics, Evaluation, Languages and literature, Literature studies, Natural Language Processing, Natural language processing, Product evaluation, bold, other},
}

Die Präsentation beschreibt die Ergebnisse verschiedener Evaluationen im Kontext des DFG-Projekts Daidalos. Dazu zählen Umfragen, Interviews und Echtzeit-Überwachung von Interaktionen mit der Projekt-Software. Mit diesen Instrumenten wurden sowohl die vom Projekt durchgeführten Workshops als auch die entwickelte Forschungsinfrastruktur (Web-Anwendung) bewertet.

@phdthesis{kotschka_grenzen_2025,
	address = {Berlin},
	type = {Masterarbeit},
	title = {Die {Grenzen} des {Wissens} für {KI}-{Modelle}},
	language = {Deutsch},
	school = {Humboldt-Universität zu Berlin},
	author = {Kotschka, Florian},
	month = sep,
	year = {2025},
	note = {im Druck},
	keywords = {other},
}

@article{shang_how_2025,
	title = {How does digital humanities research talk about {AI}? {A} bibliometric analysis},
	volume = {30},
	copyright = {Copyright (c) 2025 Wenyi Shang, Rongqian Ma, Heather Moulaison-Sandy},
	issn = {1368-1613},
	shorttitle = {How does digital humanities research talk about {AI}?},
	url = {https://publicera.kb.se/ir/article/view/47242},
	doi = {10.47989/ir30iConf47242},
	abstract = {Introduction. Despite scholarly work that looked into artificial intelligence (AI)’s impact on digital humanities (DH), there has been little comprehensive investigation into how AI has been reflected and adopted in DH scholarship. This paper addresses this gap by asking: What patterns can be observed in AI integration in the DH scholarly discourse?
Method. A bibliometric analysis was conducted on 2,488 abstracts of articles published in three influential DH journals: digital scholarship in the humanities, digital humanities quarterly, and journal of cultural analytics.
Analysis. Focusing on abstracts containing the term ‘AI’ or ‘artificial intelligence,’ this study conducted four analyses: (1) temporal analysis of AI mentions, (2) collocation analysis, (3) word vector analysis, and (4) topic modeling.
Results. AI has been discussed in DH scholarship well before its recent rise in prominence. The recurring themes in these discussions encompass both technological and human-centred aspects.
Conclusions. AI has been integral to DH, and the discourse around AI in DH reflects the field’s dual focus on technology and the humanities. The bibliometric analysis presented in this paper illustrates how information science can inform and guide methodological reflections in DH, offering new insights into the future development of the field.},
	language = {en},
	number = {iConf},
	urldate = {2025-12-01},
	journal = {Information Research. An international electronic journal},
	author = {Shang, Wenyi and Ma, Rongqian and Moulaison-Sandy, Heather},
	month = mar,
	year = {2025},
	keywords = {artificial intelligence, bibliometrics analysis, digital humanities},
	pages = {635--645},
}

Introduction. Despite scholarly work that looked into artificial intelligence (AI)’s impact on digital humanities (DH), there has been little comprehensive investigation into how AI has been reflected and adopted in DH scholarship. This paper addresses this gap by asking: What patterns can be observed in AI integration in the DH scholarly discourse? Method. A bibliometric analysis was conducted on 2,488 abstracts of articles published in three influential DH journals: digital scholarship in the humanities, digital humanities quarterly, and journal of cultural analytics. Analysis. Focusing on abstracts containing the term ‘AI’ or ‘artificial intelligence,’ this study conducted four analyses: (1) temporal analysis of AI mentions, (2) collocation analysis, (3) word vector analysis, and (4) topic modeling. Results. AI has been discussed in DH scholarship well before its recent rise in prominence. The recurring themes in these discussions encompass both technological and human-centred aspects. Conclusions. AI has been integral to DH, and the discourse around AI in DH reflects the field’s dual focus on technology and the humanities. The bibliometric analysis presented in this paper illustrates how information science can inform and guide methodological reflections in DH, offering new insights into the future development of the field.

@article{bogers_understanding_2025,
	title = {Understanding complex casual leisure information needs: an analysis of search requests for books, games, movies and music},
	volume = {81},
	issn = {0022-0418},
	url = {https://www.emerald.com/jd/article-pdf/81/1/168/9678763/jd-03-2024-0070.pdf},
	doi = {10.1108/JD-03-2024-0070},
	number = {1},
	journal = {Journal of Documentation},
	publisher = {Emerald Publishing Limited},
	author = {Bogers, Toine and Gäde, Maria and Koolen, Marijn and Petras, Vivien and Skov, Mette},
	year = {2025},
	pages = {168--194},
}

@misc{chen_reasoning_2025,
	title = {Reasoning {Models} {Don}'t {Always} {Say} {What} {They} {Think}},
	url = {https://assets.anthropic.com/m/71876fabef0f0ed4/original/reasoning_models_paper.pdf},
	publisher = {arXiv},
	author = {Chen, Yanda and Benton, Joe and Radhakrishnan, Ansh and Uesato, Jonathan and Denison, Carson and Schulman, John and Somani, Arushi and Hase, Peter and Wagner, Misha and Roger, Fabien},
	year = {2025},
}

@article{shadrova_step_2025,
	title = {'{Step} away from the {Computer}!': Über die linguistische {Datenkategorisierung} als {Erkenntnisprozess} und daraus folgende {Herausforderungen} bei der {Nachnutzung} von {Annotationen} und {Annotationstools}},
	volume = {53},
	copyright = {De Gruyter expressly reserves the right to use all content for commercial text and data mining within the meaning of Section 44b of the German Copyright Act.},
	issn = {1613-0626},
	shorttitle = {„{Step} away from the {Computer}!“},
	url = {https://www.degruyterbrill.com/document/doi/10.1515/zgl-2025-2005/html},
	doi = {10.1515/zgl-2025-2005},
	abstract = {Linguistic research frequently requires the categorization of language phenomena in corpus data (annotation). Since those may occur plentifully, a partial or full automation of the annotation process appears attractive. The filtering and recombination of existing annotation layers seems to further provide an elegant solution to the deduction of higher-level annotations. In this contribution, we show at the example of German split particle verbs that this approach results in a number of linguistic, technological, and epistemological challenges related to the precise definition of the various models employed and their interfaces. We argue that the manual annotation of corpus data is not merely a preprocessing task, but is itself an epistemological process central to the development of linguistic theory. We discuss why machine-based language processing can neither mimic nor replace this process; why it can generally not reach a level of precision that would be suitable for linguistic research without further integration, adaptation, and manual correction; and how its blind application systematically skews results in crucial areas of research. We close with the suggestion of several best practice approaches which help to prevent and resolve incompatibilities and delays arising from common problems of corpus-based language modeling.},
	language = {de},
	number = {1},
	urldate = {2025-11-17},
	journal = {Zeitschrift für germanistische Linguistik},
	publisher = {De Gruyter},
	author = {Shadrova, Anna and Lüdeling, Anke and Klotz, Martin and Hartz, Rahel Gajaneh and Krause, Thomas},
	month = apr,
	year = {2025},
	pages = {166--214},
}

Linguistic research frequently requires the categorization of language phenomena in corpus data (annotation). Since those may occur plentifully, a partial or full automation of the annotation process appears attractive. The filtering and recombination of existing annotation layers seems to further provide an elegant solution to the deduction of higher-level annotations. In this contribution, we show at the example of German split particle verbs that this approach results in a number of linguistic, technological, and epistemological challenges related to the precise definition of the various models employed and their interfaces. We argue that the manual annotation of corpus data is not merely a preprocessing task, but is itself an epistemological process central to the development of linguistic theory. We discuss why machine-based language processing can neither mimic nor replace this process; why it can generally not reach a level of precision that would be suitable for linguistic research without further integration, adaptation, and manual correction; and how its blind application systematically skews results in crucial areas of research. We close with the suggestion of several best practice approaches which help to prevent and resolve incompatibilities and delays arising from common problems of corpus-based language modeling.

@techreport{rat_fur_informationsinfrastrukturen_leistung_2025,
	title = {Leistung in {Verantwortung}. {Zur} {Zukunft} der wissenschaftlichen {Informationsinfrastrukturen} in {Deutschland}},
	url = {https://rfii.de/download/leistung-in-verantwortung-juli-2025/},
	abstract = {Positionspapier Leistung in Verantwortung – Juli 2025 Der Rat für Informationsinfrastrukturen (RfII) wendet sich an die Wissenschaftspolitik sowie an wissenschaftliche […]},
	language = {de-DE},
	urldate = {2025-09-30},
	author = {Rat für Informationsinfrastrukturen},
	year = {2025},
	pages = {61},
}

Positionspapier Leistung in Verantwortung – Juli 2025 Der Rat für Informationsinfrastrukturen (RfII) wendet sich an die Wissenschaftspolitik sowie an wissenschaftliche […]

@inproceedings{agarwal_exploring_2025,
	address = {Cham},
	title = {Exploring the {Use} of {Educational} {Data} {Mining} and {Learning} {Analytics} {Through} {AI} to {Improve} {Instructional} {Practices} and {Student} {Performance}},
	isbn = {978-3-031-84394-5},
	doi = {10.1007/978-3-031-84394-5_6},
	abstract = {AI learning analytics (AI LA) and educational data mining (EDM) are two new fields that use the power of data analysis to better educational practises and student performance. These approaches give teachers access to information about the behaviour, learning patterns, and performance of their students by analysing big datasets gathered from diverse educational resources. Personalising learning experiences, offering early interventions and assistance, informing curriculum and instructional design, utilising predictive analytics for preventative measures, improving assessment and feedback systems, and guiding institutional decision-making are all possible uses for this information. To achieve proper and efficient implementation, ethical issues like prejudice, consent, and data protection must be properly considered. Overall, educational data mining and AI Learning analytics have enormous potential to alter education and give teachers the tools they need to optimise teaching learning Process.},
	language = {en},
	booktitle = {Artificial {Intelligence} and {Its} {Applications}},
	publisher = {Springer Nature Switzerland},
	author = {Agarwal, Nidhi and Babu, Yogendra and Awadh, Ram and Mishra, Vikas},
	editor = {Gupta, Anish and Hinchey, Michael and Zalevsky, Zeev},
	year = {2025},
	keywords = {AI Learning analytics, Educational Data Mining, Instructional Practices, Performance},
	pages = {60--71},
}

AI learning analytics (AI LA) and educational data mining (EDM) are two new fields that use the power of data analysis to better educational practises and student performance. These approaches give teachers access to information about the behaviour, learning patterns, and performance of their students by analysing big datasets gathered from diverse educational resources. Personalising learning experiences, offering early interventions and assistance, informing curriculum and instructional design, utilising predictive analytics for preventative measures, improving assessment and feedback systems, and guiding institutional decision-making are all possible uses for this information. To achieve proper and efficient implementation, ethical issues like prejudice, consent, and data protection must be properly considered. Overall, educational data mining and AI Learning analytics have enormous potential to alter education and give teachers the tools they need to optimise teaching learning Process.

@misc{skorinkin_quantitative_2025,
	title = {Quantitative {Analyse} der {Medienwellen} der {Spanischen} {Grippe} (1918/19). {Eine} {Fallstudie}},
	url = {https://zenodo.org/records/15703503},
	doi = {10.5281/zenodo.15703503},
	abstract = {Die vorliegende Fallstudie bereitet – in Form eines "Jupyter Books" – den Prozess und die Ergebnisse eines Forschungsprojekts aus den Digital Humanities didaktisch auf.},
	urldate = {2025-10-05},
	publisher = {Zenodo},
	author = {Skorinkin, Daniil and Sluyter-Gäthje, Henny and Trilcke, Peer},
	month = jun,
	year = {2025},
	keywords = {open educational resource},
}

Die vorliegende Fallstudie bereitet – in Form eines "Jupyter Books" – den Prozess und die Ergebnisse eines Forschungsprojekts aus den Digital Humanities didaktisch auf.

@misc{schulz_responsible_2025,
	address = {Berlin},
	title = {Responsible {AI} \& {Dokumentation}},
	url = {https://zenodo.org/records/14716501},
	doi = {10.5281/zenodo.14716500},
	abstract = {Diese Präsentation widmet sich der verantwortungsvollen Entwicklung und Dokumentation von KI-Systemen. Sie bietet einen Überblick über zentrale ethische Herausforderungen und deren Bedeutung in der Praxis. Unter Berücksichtigung aktueller Forschungsarbeiten werden Schlüsselthemen wie Werteorientierung bei Datensätzen, Bias-Erkennung sowie die Integration ethischer Normen im Entwicklungsprozess beleuchtet. Ergänzt wird dies durch Ansätze wie das "Red Teaming" von Sprachmodellen, um potenzielle Schwachstellen offenzulegen.

Ein weiterer Fokus liegt auf den Anforderungen an die Dokumentation von KI-Systemen gemäß EU AI Act, einschließlich Aspekten wie Transparenz, menschlicher Aufsicht, Cybersicherheit und Risikoanalyse. Die Präsentation veranschaulicht auch die Initiativen der Humboldt-Universität zu Berlin, darunter Leitlinien zur Nutzung generativer KI in Forschung und Lehre sowie deren Einbettung in eine breite Governance-Strategie.

Abschließend werden praktische Instrumente wie Model Cards und Data Statements vorgestellt, die die strukturierte Dokumentation von KI-Systemen fördern. Diese Ansätze bieten eine Grundlage für ethisch und technisch fundierte KI-Lösungen und unterstützen die Schaffung eines öffentlichen Transparenzstandards, wie er in Städten wie Helsinki und Amsterdam umgesetzt wird.},
	language = {deu},
	urldate = {2025-01-22},
	author = {Schulz, Konstantin},
	month = jan,
	year = {2025},
	keywords = {Artificial Intelligence, Artificial intelligence, Natural Language Processing, Natural language processing, other},
}

Diese Präsentation widmet sich der verantwortungsvollen Entwicklung und Dokumentation von KI-Systemen. Sie bietet einen Überblick über zentrale ethische Herausforderungen und deren Bedeutung in der Praxis. Unter Berücksichtigung aktueller Forschungsarbeiten werden Schlüsselthemen wie Werteorientierung bei Datensätzen, Bias-Erkennung sowie die Integration ethischer Normen im Entwicklungsprozess beleuchtet. Ergänzt wird dies durch Ansätze wie das "Red Teaming" von Sprachmodellen, um potenzielle Schwachstellen offenzulegen. Ein weiterer Fokus liegt auf den Anforderungen an die Dokumentation von KI-Systemen gemäß EU AI Act, einschließlich Aspekten wie Transparenz, menschlicher Aufsicht, Cybersicherheit und Risikoanalyse. Die Präsentation veranschaulicht auch die Initiativen der Humboldt-Universität zu Berlin, darunter Leitlinien zur Nutzung generativer KI in Forschung und Lehre sowie deren Einbettung in eine breite Governance-Strategie. Abschließend werden praktische Instrumente wie Model Cards und Data Statements vorgestellt, die die strukturierte Dokumentation von KI-Systemen fördern. Diese Ansätze bieten eine Grundlage für ethisch und technisch fundierte KI-Lösungen und unterstützen die Schaffung eines öffentlichen Transparenzstandards, wie er in Städten wie Helsinki und Amsterdam umgesetzt wird.

@misc{noauthor_2024_06_berlin03_embeddings_demoipynb_2025,
	title = {2024\_06\_Berlin/03\_embeddings\_demo.ipynb · main · daidalos / daidalos-workshops · {GitLab}},
	url = {https://scm.cms.hu-berlin.de/daidalos/daidalos-workshops/-/blob/main/2024_06_Berlin/03_embeddings_demo.ipynb?ref_type=heads},
	abstract = {GitLab - Software zur Zusammenarbeit an Quellcode},
	language = {en},
	urldate = {2025-07-07},
	journal = {GitLab},
	month = jan,
	year = {2025},
}

GitLab - Software zur Zusammenarbeit an Quellcode

@misc{beyer_llm_2025,
	title = {{LLM} \& {NLP} in der {Klassischen} {Philologie}: {Digitale} {Literaturwissenschaft} für alle?},
	shorttitle = {{LLM} \& {NLP} in der {Klassischen} {Philologie}},
	url = {https://zenodo.org/records/14679224},
	doi = {10.5281/zenodo.14679224},
	abstract = {Die Präsentation "LLM \& NLP in der Klassischen Philologie: Digitale Literaturwissenschaft für alle?" beleuchtet den Einsatz von Natural Language Processing (NLP) und Large Language Models (LLMs) in der Analyse antiker Texte, mit besonderem Fokus auf Sallusts Monographien Bellum Catilinae und Bellum Iugurthinum. Im Zentrum steht die Frage nach den emotionalen Nuancen der Protagonisten Catilina und Iugurtha, die durch Sentiment-Analysen untersucht werden.

Durch die Kombination moderner NLP-Methoden und spezialisierter LLMs wie GPT-4 und Gemini-1.5-Pro wird eine Polaritätsbewertung der Texte vorgenommen. Neben der Vorstellung lexikonbasierter Ansätze und Limitierungen wird die Entwicklung spezifischer Prompting-Strategien für LLMs diskutiert. Methodische Herausforderungen wie geringe Sentiment-Quote und Unterschiede in der Modellarchitektur werden thematisiert, ebenso wie Chancen durch Multimodalität und datenbasierte Erweiterungen.

Die Ergebnisse liefern Einblicke in die emotionale Charakterisierung und eröffnen neue Perspektiven für die digitale Literaturwissenschaft. Abschließend wird die Zukunft digitaler Methoden in den Geisteswissenschaften erörtert, einschließlich der Bedeutung interdisziplinärer Zusammenarbeit und der Nutzung von Retrieval-Augmented Generation (RAG) sowie Low-Rank Adaptation (LoRA) für historische Sprachen.},
	language = {deu},
	urldate = {2025-01-22},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jan,
	year = {2025},
	keywords = {Artificial Intelligence, Artificial intelligence, Classics, Languages and literature, Latin, Literature, Literature studies, Literature study, Natural Language Processing, Natural language processing},
}

Die Präsentation "LLM & NLP in der Klassischen Philologie: Digitale Literaturwissenschaft für alle?" beleuchtet den Einsatz von Natural Language Processing (NLP) und Large Language Models (LLMs) in der Analyse antiker Texte, mit besonderem Fokus auf Sallusts Monographien Bellum Catilinae und Bellum Iugurthinum. Im Zentrum steht die Frage nach den emotionalen Nuancen der Protagonisten Catilina und Iugurtha, die durch Sentiment-Analysen untersucht werden. Durch die Kombination moderner NLP-Methoden und spezialisierter LLMs wie GPT-4 und Gemini-1.5-Pro wird eine Polaritätsbewertung der Texte vorgenommen. Neben der Vorstellung lexikonbasierter Ansätze und Limitierungen wird die Entwicklung spezifischer Prompting-Strategien für LLMs diskutiert. Methodische Herausforderungen wie geringe Sentiment-Quote und Unterschiede in der Modellarchitektur werden thematisiert, ebenso wie Chancen durch Multimodalität und datenbasierte Erweiterungen. Die Ergebnisse liefern Einblicke in die emotionale Charakterisierung und eröffnen neue Perspektiven für die digitale Literaturwissenschaft. Abschließend wird die Zukunft digitaler Methoden in den Geisteswissenschaften erörtert, einschließlich der Bedeutung interdisziplinärer Zusammenarbeit und der Nutzung von Retrieval-Augmented Generation (RAG) sowie Low-Rank Adaptation (LoRA) für historische Sprachen.

@misc{liao_llms_2024,
	title = {{LLMs} as {Research} {Tools}: {A} {Large} {Scale} {Survey} of {Researchers}' {Usage} and {Perceptions}},
	shorttitle = {{LLMs} as {Research} {Tools}},
	url = {http://arxiv.org/abs/2411.05025},
	doi = {10.48550/arXiv.2411.05025},
	abstract = {The rise of large language models (LLMs) has led many researchers to consider their usage for scientific work. Some have found benefits using LLMs to augment or automate aspects of their research pipeline, while others have urged caution due to risks and ethical concerns. Yet little work has sought to quantify and characterize how researchers use LLMs and why. We present the first large-scale survey of 816 verified research article authors to understand how the research community leverages and perceives LLMs as research tools. We examine participants' self-reported LLM usage, finding that 81\% of researchers have already incorporated LLMs into different aspects of their research workflow. We also find that traditionally disadvantaged groups in academia (non-White, junior, and non-native English speaking researchers) report higher LLM usage and perceived benefits, suggesting potential for improved research equity. However, women, non-binary, and senior researchers have greater ethical concerns, potentially hindering adoption.},
	urldate = {2026-02-05},
	publisher = {arXiv},
	author = {Liao, Zhehui and Antoniak, Maria and Cheong, Inyoung and Cheng, Evie Yu-Yen and Lee, Ai-Heng and Lo, Kyle and Chang, Joseph Chee and Zhang, Amy X.},
	month = oct,
	year = {2024},
	note = {arXiv:2411.05025 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Computers and Society, Computer Science - Digital Libraries, Computer Science - Human-Computer Interaction},
}

The rise of large language models (LLMs) has led many researchers to consider their usage for scientific work. Some have found benefits using LLMs to augment or automate aspects of their research pipeline, while others have urged caution due to risks and ethical concerns. Yet little work has sought to quantify and characterize how researchers use LLMs and why. We present the first large-scale survey of 816 verified research article authors to understand how the research community leverages and perceives LLMs as research tools. We examine participants' self-reported LLM usage, finding that 81% of researchers have already incorporated LLMs into different aspects of their research workflow. We also find that traditionally disadvantaged groups in academia (non-White, junior, and non-native English speaking researchers) report higher LLM usage and perceived benefits, suggesting potential for improved research equity. However, women, non-binary, and senior researchers have greater ethical concerns, potentially hindering adoption.

@incollection{reinhard_augmentierung_2024,
	address = {Wiesbaden},
	title = {Augmentierung statt {Automatisierung}: {Nachhaltigkeit} in der {Entwicklung} von {KI}-{Systemen} aus der {Mitarbeitendenperspektive}},
	isbn = {978-3-658-45148-6},
	shorttitle = {Augmentierung statt {Automatisierung}},
	url = {https://doi.org/10.1007/978-3-658-45148-6_15},
	doi = {10.1007/978-3-658-45148-6_15},
	abstract = {In diesem Beitrag wird durch eine systematische Literaturrecherche das Potential der KI zur Augmentierung im Servicebereich, insbesondere im Kundenservice, beleuchtet. Angesichts der steigenden Herausforderungen für Servicemitarbeitende zielt unsere Untersuchung darauf ab, KI nicht als Ersatz, sondern als Unterstützung für menschliche Fähigkeiten zu positionieren. Trotz der bestehenden Herausforderungen und Grenzen der KI in komplexen, kontextabhängigen Umgebungen, wie dem Kundenservice, verdeutlicht unsere Forschung, wie KI-Systeme Arbeitsprozesse verbessern können, indem sie Entscheidungsprozesse unterstützen, Zugang zu Kundendaten gewähren und wertvolles Feedback liefern. Dieser Beitrag schließt eine wichtige Forschungslücke, indem er aufzeigt, wie eine nachhaltige und verantwortungsvolle Integration von KI in Servicesysteme zur Steigerung der Servicequalität und Mitarbeiterzufriedenheit beitragen kann.},
	language = {de},
	urldate = {2025-12-30},
	booktitle = {Sustainable {Service} {Management}: {Band} 1: {Nachhaltigkeit} aus {Sicht} von {Kunden} und {Mitarbeitern}},
	publisher = {Springer Fachmedien},
	author = {Reinhard, Philipp and Li, Mahei Manhai and Peters, Christoph and Leimeister, Jan Marco},
	editor = {Bruhn, Manfred and Hadwich, Karsten},
	year = {2024},
	pages = {465--492},
}

In diesem Beitrag wird durch eine systematische Literaturrecherche das Potential der KI zur Augmentierung im Servicebereich, insbesondere im Kundenservice, beleuchtet. Angesichts der steigenden Herausforderungen für Servicemitarbeitende zielt unsere Untersuchung darauf ab, KI nicht als Ersatz, sondern als Unterstützung für menschliche Fähigkeiten zu positionieren. Trotz der bestehenden Herausforderungen und Grenzen der KI in komplexen, kontextabhängigen Umgebungen, wie dem Kundenservice, verdeutlicht unsere Forschung, wie KI-Systeme Arbeitsprozesse verbessern können, indem sie Entscheidungsprozesse unterstützen, Zugang zu Kundendaten gewähren und wertvolles Feedback liefern. Dieser Beitrag schließt eine wichtige Forschungslücke, indem er aufzeigt, wie eine nachhaltige und verantwortungsvolle Integration von KI in Servicesysteme zur Steigerung der Servicequalität und Mitarbeiterzufriedenheit beitragen kann.

@inproceedings{natario_texeira_augmented_2024,
	address = {New York, NY, USA},
	title = {Augmented {Reading}},
	url = {https://dl.acm.org/doi/10.1145/3604479.3604511},
	doi = {https://doi.org/10.1145/3604479.3604511},
	abstract = {Reading is one of the basic activities inherent to humans. By reading, we communicate to each other, sharing and receiving information. In a world technology is constantly evolving, it is not hard to find software solutions for speed reading. However, the true benefit of some of those solutions is still unclear. In this work, we verify the viability of implementing AR-based speed reading solutions, so that one may use them with real books, instead of digital ones. We implement an AR equivalent of two well-known solutions, Bionic Reading and Spritz, show the perceived limitations and confirm that AR can also try to improve reading tasks.},
	language = {EN},
	urldate = {2025-12-30},
	booktitle = {Proceedings of the 24th {Symposium} on {Virtual} and {Augmented} {Reality}},
	publisher = {Association for Computing Machinery},
	author = {Natario Texeira, João Marcelo Xavier and Dias Tavares de Souza, Júlia},
	year = {2024},
	pages = {55--61},
}

Reading is one of the basic activities inherent to humans. By reading, we communicate to each other, sharing and receiving information. In a world technology is constantly evolving, it is not hard to find software solutions for speed reading. However, the true benefit of some of those solutions is still unclear. In this work, we verify the viability of implementing AR-based speed reading solutions, so that one may use them with real books, instead of digital ones. We implement an AR equivalent of two well-known solutions, Bionic Reading and Spritz, show the perceived limitations and confirm that AR can also try to improve reading tasks.

@inproceedings{hanel_towards_2024,
	title = {Towards {Effective} {AI}-{Driven} {Reading} {Assistants}: {A} {Design} {Science} {Exploration}},
	shorttitle = {Towards {Effective} {AI}-{Driven} {Reading} {Assistants}},
	url = {https://aisel.aisnet.org/ecis2024/track23_designresearch/track23_designresearch/12},
	abstract = {Recent advancements in AI have led to the introduction of tools that support researchers in scientific reading. Tools such as SciSpace have come to the forefront to assist users in reading scientific texts. However, there is an insufficient theoretical foundation on how to design these reading assistants as well as no evidence of their effects, especially given the recent progress. Specifically, past literature lacks insights on evaluated user requirements and design principles for the design of computer-assisted reading systems. Addressing these challenges, we draw on Design Science Research (DSR) to derive and evaluate a set of five design principles for computer-assisted reading systems. Building on flow theory as our theoretical lens, we develop and perform a first proof-of-concept evaluation of a prototypical implementation of our principles as a computer-assisted reading artifact. Our design principles support researchers and practitioners on how to design, evaluate, and compare their AI-reading tools more effectively.},
	booktitle = {{ECIS} 2024 {Proceedings}: {People} {First}: {Constructing} {Digital} {Futures} {Together}},
	publisher = {Association of Information Systems in Region 2},
	author = {Hänel, Martin and Wambsganss, Thiemo and Söllner, Matthias},
	month = jun,
	year = {2024},
}

Recent advancements in AI have led to the introduction of tools that support researchers in scientific reading. Tools such as SciSpace have come to the forefront to assist users in reading scientific texts. However, there is an insufficient theoretical foundation on how to design these reading assistants as well as no evidence of their effects, especially given the recent progress. Specifically, past literature lacks insights on evaluated user requirements and design principles for the design of computer-assisted reading systems. Addressing these challenges, we draw on Design Science Research (DSR) to derive and evaluate a set of five design principles for computer-assisted reading systems. Building on flow theory as our theoretical lens, we develop and perform a first proof-of-concept evaluation of a prototypical implementation of our principles as a computer-assisted reading artifact. Our design principles support researchers and practitioners on how to design, evaluate, and compare their AI-reading tools more effectively.

@inproceedings{ribeiro_digital_2024,
	address = {New York, NY, USA},
	title = {Digital {Augmentation} in {Special} {Needs} {Reading}: {Enhancing} {Inclusiveness}},
	shorttitle = {Digital {Augmentation} in {Special} {Needs} {Reading}},
	url = {https://dl.acm.org/doi/10.1145/3628516.3659405},
	doi = {https://doi.org/10.1145/3628516.3659405},
	abstract = {In the evolving landscape of educational technology, Digital Augmentation of Reading, incorporating digital media augmentations, including sound, illustrations and light effects, is being explored to enhance reading comprehension, motivation, and overall learning experiences. This paper presents an ongoing initiative in its early stages, focusing on optimizing reading instruction for special needs education. Leveraging an established design approach– rooted in Design-Based Research, Design Space exploration, and Design Thinking–that has previously proven successful in creating STREEN, a digitally augmented reading technology aimed at promoting reading comprehension and motivation among third-grade students. With this approach, we aim to redesign STREEN for special needs education. Utilizing insights from a previously published study on needs in a school for students with special needs, we engaged with school stakeholders to envision the repurposing of STREEN. We propose evaluation activities to assess teachers’ acceptance of STREEN and measure its impact on student engagement. In addition, we established a Design Thinking Club in a primary school to support the generation of solutions for the adaptation of STREEN. Finally, we developed a Virtual Reality Simulator for rapid prototyping of STREEN.},
	language = {EN},
	urldate = {2025-12-30},
	booktitle = {Proceedings of the 23rd {Annual} {ACM} {Interaction} {Design} and {Children} {Conference}},
	publisher = {Association for Computing Machinery},
	author = {Ribeiro, Pedro and Wild-Wall, Nele and Ressel, Christian},
	year = {2024},
	doi = {10.1145/3628516.3659405},
	pages = {844--849},
}

In the evolving landscape of educational technology, Digital Augmentation of Reading, incorporating digital media augmentations, including sound, illustrations and light effects, is being explored to enhance reading comprehension, motivation, and overall learning experiences. This paper presents an ongoing initiative in its early stages, focusing on optimizing reading instruction for special needs education. Leveraging an established design approach– rooted in Design-Based Research, Design Space exploration, and Design Thinking–that has previously proven successful in creating STREEN, a digitally augmented reading technology aimed at promoting reading comprehension and motivation among third-grade students. With this approach, we aim to redesign STREEN for special needs education. Utilizing insights from a previously published study on needs in a school for students with special needs, we engaged with school stakeholders to envision the repurposing of STREEN. We propose evaluation activities to assess teachers’ acceptance of STREEN and measure its impact on student engagement. In addition, we established a Design Thinking Club in a primary school to support the generation of solutions for the adaptation of STREEN. Finally, we developed a Virtual Reality Simulator for rapid prototyping of STREEN.

@article{lo_semantic_2024,
	title = {The {Semantic} {Reader} {Project}: {Augmenting} {Scholarly} {Documents} through {AI}-{Powered} {Interactive} {Reading} {Interfaces}},
	volume = {67},
	shorttitle = {The {Semantic} {Reader} {Project}},
	url = {https://dl.acm.org/doi/pdf/10.1145/3659096},
	doi = {https://dx.doi.org/10.1145/3659096},
	abstract = {The exponential growth in the rate of scientific
publication and increasing interdisciplinary nature
of scientific progress makes it increasingly hard for scholars to keep up with the latest developments. Academic search engines, such as Google Scholar and Semantic Scholar, help scholars discover research papers. Techniques such as automated summarization help scholars triage research papers. But when it comes to actually reading research papers, the process,
often based on a static PDF format, has remained
largely unchanged for many decades. This is a problem because digesting technical research papers in their conventional formats is difficult.},
	language = {en},
	number = {10},
	urldate = {2025-12-30},
	journal = {Communications of the ACM},
	publisher = {arXiv},
	author = {Lo, Kyle and Chang, Joseph Chee and Head, Andrew and Bragg, Jonathan and Zhang, Amy X. and Trier, Cassidy and Anastasiades, Chloe and August, Tal and Authur, Russell and Bragg, Danielle and + 45 authors and Hearst, Marti A. and Weld, Daniel S.},
	year = {2024},
	note = {https://scholarphi.org/},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Human-Computer Interaction},
	pages = {50--61},
}

The exponential growth in the rate of scientific publication and increasing interdisciplinary nature of scientific progress makes it increasingly hard for scholars to keep up with the latest developments. Academic search engines, such as Google Scholar and Semantic Scholar, help scholars discover research papers. Techniques such as automated summarization help scholars triage research papers. But when it comes to actually reading research papers, the process, often based on a static PDF format, has remained largely unchanged for many decades. This is a problem because digesting technical research papers in their conventional formats is difficult.

@mastersthesis{bell_augmented_2024,
	address = {Wellington},
	title = {Augmented {Text}: {A} {New} {Frontier} in {Comprehension}},
	url = {https://openaccess.wgtn.ac.nz/articles/thesis/Augmented_Text_A_New_Frontier_in_Comprehension/27020665?file=49188916},
	urldate = {2025-12-30},
	school = {Te Herenga Waka—Victoria University of Wellington},
	author = {Bell, Jon},
	month = sep,
	year = {2024},
	note = {https://www.augmented-text.com/thesis},
}

@misc{fu_why_2024,
	title = {Why {Do} {Large} {Language} {Models} ({LLMs}) {Struggle} to {Count} {Letters}?},
	url = {http://arxiv.org/abs/2412.18626},
	doi = {10.48550/arXiv.2412.18626},
	abstract = {TAIRAN FU, College of Mechanical \& Electrical Engineering, Nanjing University of Aeronautics and Astronautics, China RAQUEL FERRANDO, JAVIER CONDE, CARLOS ARRIAGA, and PEDRO REVIRIEGO, ETSI de Telecomunicación, Universidad Politécnica de Madrid, Spain Large Language Models (LLMs) have achieved unprecedented performance on many complex tasks, being able, for example, to answer questions on almost any topic. However, they struggle with other simple tasks, such as counting the occurrences of letters in a word, as illustrated by the inability of many LLMs to count the number of "r" letters in "strawberry". Several works have studied this problem and linked it to the tokenization used by LLMs, to the intrinsic limitations of the attention mechanism, or to the lack of character-level training data. In this paper, we conduct an experimental study to evaluate the relations between the LLM errors when counting letters with 1) the frequency of the word and its components in the training dataset and 2) the complexity of the counting operation. We present a comprehensive analysis of the errors of LLMs when counting letter occurrences by evaluating a representative group of models over a large number of words. The results show a number of consistent trends in the models evaluated: 1) models are capable of recognizing the letters but not counting them; 2) the frequency of the word and tokens in the word does not have a significant impact on the LLM errors; 3) there is a positive correlation of letter frequency with errors, more frequent letters tend to have more counting errors, 4) the errors show a strong correlation with the number of letters or tokens in a word and 5) the strongest correlation occurs with the number of letters with counts larger than one, with most models being unable to correctly count words in which letters appear more than twice. These results suggest that the problems of LLMs to count letters are not related to the frequency of words or tokens in the training data but to the complexity of the counting operation. However, further studies are needed to build a better understanding of the limitations of LLMs to count the letters in a word. CCS Concepts: • Software and its engineering → Empirical software validation; • Computing methodologies → Natural language generation; Language resources.},
	language = {en},
	urldate = {2025-12-28},
	publisher = {arXiv},
	author = {Fu, Tairan and Ferrando, Raquel and Conde, Javier and Arriaga, Carlos and Reviriego, Pedro},
	month = dec,
	year = {2024},
	note = {arXiv:2412.18626 [cs]},
	keywords = {Computer Science - Computation and Language},
}

TAIRAN FU, College of Mechanical & Electrical Engineering, Nanjing University of Aeronautics and Astronautics, China RAQUEL FERRANDO, JAVIER CONDE, CARLOS ARRIAGA, and PEDRO REVIRIEGO, ETSI de Telecomunicación, Universidad Politécnica de Madrid, Spain Large Language Models (LLMs) have achieved unprecedented performance on many complex tasks, being able, for example, to answer questions on almost any topic. However, they struggle with other simple tasks, such as counting the occurrences of letters in a word, as illustrated by the inability of many LLMs to count the number of "r" letters in "strawberry". Several works have studied this problem and linked it to the tokenization used by LLMs, to the intrinsic limitations of the attention mechanism, or to the lack of character-level training data. In this paper, we conduct an experimental study to evaluate the relations between the LLM errors when counting letters with 1) the frequency of the word and its components in the training dataset and 2) the complexity of the counting operation. We present a comprehensive analysis of the errors of LLMs when counting letter occurrences by evaluating a representative group of models over a large number of words. The results show a number of consistent trends in the models evaluated: 1) models are capable of recognizing the letters but not counting them; 2) the frequency of the word and tokens in the word does not have a significant impact on the LLM errors; 3) there is a positive correlation of letter frequency with errors, more frequent letters tend to have more counting errors, 4) the errors show a strong correlation with the number of letters or tokens in a word and 5) the strongest correlation occurs with the number of letters with counts larger than one, with most models being unable to correctly count words in which letters appear more than twice. These results suggest that the problems of LLMs to count letters are not related to the frequency of words or tokens in the training data but to the complexity of the counting operation. However, further studies are needed to build a better understanding of the limitations of LLMs to count the letters in a word. CCS Concepts: • Software and its engineering → Empirical software validation; • Computing methodologies → Natural language generation; Language resources.

@misc{beyer_can_2024,
	address = {Athens},
	title = {Can {Jupyter} {Help} {Daidalos}? {Or}: {How} to {Develop} {Digital} {Literacies} and {Assess} {Them}?},
	shorttitle = {Can {Jupyter} {Help} {Daidalos}?},
	url = {https://zenodo.org/records/12200296},
	doi = {10.5281/zenodo.12200296},
	abstract = {Das Daidalos-Projekt entwickelt eine benutzerzentrierte NLP-Infrastruktur für Forschende der Klassischen Philologie und verwandter Disziplinen, um den Einsatz von Natural Language Processing (NLP) in der Erforschung lateinischer und griechischer Textkorpora zu fördern. Ziel ist es, Forschende mit unterschiedlichen digitalen Kompetenzen durch eine interaktive Plattform zu unterstützen, die verschiedene NLP-Methoden wie Named Entity Recognition (NER), Part-of-Speech-Tagging und Sentiment-Analyse bereitstellt. Die Plattform wird auf der Basis von Forschungstandems entwickelt, in denen Forschungsfragen, Textkorpora und NLP-Methoden gemeinsam erarbeitet werden. Im Zentrum steht die Entwicklung eines domain-spezifischen Methodenrahmens für digitale Kompetenzen, das die Dimensionen Digital Literacy, Data Literacy, AI Literacy umfasst. Dieses Framework wird anhand konkreter Forschungsfragen, wie der Identifikation impliziter Ereignisse in historischen Texten, operationalisiert und in modularen Lernangeboten – insbesondere in Jupyter Notebooks – umgesetzt. Die Infrastruktur zielt darauf ab, Forschende mit fortgeschrittenen beruflichen Verpflichtungen durch praxisnahe, anwendungsorientierte Weiterbildung in iher digitalen Forschungskompetenz zu stärken und institutionelle Rahmenbedingungen für die Förderung digitaler Kompetenzen zu schaffen.},
	language = {eng},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = jun,
	year = {2024},
	keywords = {Classics, Digital Literacies, Digital humanities, Jupyter Notebooks, other},
}

Das Daidalos-Projekt entwickelt eine benutzerzentrierte NLP-Infrastruktur für Forschende der Klassischen Philologie und verwandter Disziplinen, um den Einsatz von Natural Language Processing (NLP) in der Erforschung lateinischer und griechischer Textkorpora zu fördern. Ziel ist es, Forschende mit unterschiedlichen digitalen Kompetenzen durch eine interaktive Plattform zu unterstützen, die verschiedene NLP-Methoden wie Named Entity Recognition (NER), Part-of-Speech-Tagging und Sentiment-Analyse bereitstellt. Die Plattform wird auf der Basis von Forschungstandems entwickelt, in denen Forschungsfragen, Textkorpora und NLP-Methoden gemeinsam erarbeitet werden. Im Zentrum steht die Entwicklung eines domain-spezifischen Methodenrahmens für digitale Kompetenzen, das die Dimensionen Digital Literacy, Data Literacy, AI Literacy umfasst. Dieses Framework wird anhand konkreter Forschungsfragen, wie der Identifikation impliziter Ereignisse in historischen Texten, operationalisiert und in modularen Lernangeboten – insbesondere in Jupyter Notebooks – umgesetzt. Die Infrastruktur zielt darauf ab, Forschende mit fortgeschrittenen beruflichen Verpflichtungen durch praxisnahe, anwendungsorientierte Weiterbildung in iher digitalen Forschungskompetenz zu stärken und institutionelle Rahmenbedingungen für die Förderung digitaler Kompetenzen zu schaffen.

@misc{beyer_daidalos_2024,
	address = {Leipzig},
	title = {Daidalos: {NER} for {Literary} {Studies} on {Latin} and {Ancient} {Greek} {Texts}},
	shorttitle = {Daidalos},
	url = {https://zenodo.org/records/12582628},
	doi = {10.5281/zenodo.12582628},
	abstract = {The talk presents Daidalos, an NLP infrastructure designed to support researchers in Classical Philology and related disciplines in applying natural language processing methods to Latin and Ancient Greek texts. The project addresses the unique challenges of processing ancient languages, including discontinuous, nested, and overlapping entity spans, ambiguity, and complex linguistic phenomena such as metonymy and ellipsis. Daidalos offers a platform integrating multiple NLP methods—such as Named Entity Recognition (NER), part-of-speech tagging, and sentiment analysis—alongside customizable pipelines tailored to literary research questions. The infrastructure emphasizes transparency and sustainability through model cards, data sheets, and well-documented evaluations. It supports both standalone NER applications and integration into broader research pipelines, enabling tasks like identifying implicit references in historiographical texts. In teaching, Daidalos promotes digital literacies via curated Jupyter Notebooks, model cards, and datasheets, facilitating accessible learning. The project also explores the evolving role of generative AI in Classics, questioning the necessity of traditional taggers in light of emerging large language models. Overall, Daidalos aims to foster a community of practice that combines methodological rigor with open science principles.},
	language = {eng},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = jun,
	year = {2024},
	keywords = {Classics, Literature studies, NER, NLP, other},
}

The talk presents Daidalos, an NLP infrastructure designed to support researchers in Classical Philology and related disciplines in applying natural language processing methods to Latin and Ancient Greek texts. The project addresses the unique challenges of processing ancient languages, including discontinuous, nested, and overlapping entity spans, ambiguity, and complex linguistic phenomena such as metonymy and ellipsis. Daidalos offers a platform integrating multiple NLP methods—such as Named Entity Recognition (NER), part-of-speech tagging, and sentiment analysis—alongside customizable pipelines tailored to literary research questions. The infrastructure emphasizes transparency and sustainability through model cards, data sheets, and well-documented evaluations. It supports both standalone NER applications and integration into broader research pipelines, enabling tasks like identifying implicit references in historiographical texts. In teaching, Daidalos promotes digital literacies via curated Jupyter Notebooks, model cards, and datasheets, facilitating accessible learning. The project also explores the evolving role of generative AI in Classics, questioning the necessity of traditional taggers in light of emerging large language models. Overall, Daidalos aims to foster a community of practice that combines methodological rigor with open science principles.

@misc{beyer_digital_2024,
	address = {Berlin},
	title = {Digital {Literacies}},
	url = {https://zenodo.org/records/10515036},
	doi = {10.5281/zenodo.10515036},
	abstract = {Der Vortrag untersucht die Entwicklung und Konzeption von Digital Literacies im zeitlichen Kontext, mit Fokus auf die drei zentralen Kompetenzbereiche Digital Literacy, Data Literacy und AI Literacy. Ausgehend von der ersten Erwähnung des Begriffs „Digital Literacy“ im Jahr 1997 wird die zunehmende Akademisierung und institutionelle Verankerung dieser Konzepte seit 2015, insbesondere im Bereich Data Literacy, dargestellt. Die Analyse zeigt eine signifikante Zunahme wissenschaftlicher Publikationen zu diesen Themen, was auf eine wachsende gesellschaftliche und bildungspolitische Relevanz hinweist. Dabei werden unterschiedliche Modellansätze wie das Kompetenzrahmenmodell, das Kompetenzstufenmodell sowie spezifische Frameworks wie TPACK, DPACK und AI-PACK vorgestellt, die die Integration von Fachwissen, technologischer Kompetenz und pädagogischer Praxis unterstützen. Besonderes Augenmerk gilt der transdisziplinären und lebenslangen Vermittlung von Kompetenzen, die ethische, kritische und reflektierte Auseinandersetzung mit Daten und Künstlicher Intelligenz einschließen.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = jan,
	year = {2024},
	keywords = {other},
}

Der Vortrag untersucht die Entwicklung und Konzeption von Digital Literacies im zeitlichen Kontext, mit Fokus auf die drei zentralen Kompetenzbereiche Digital Literacy, Data Literacy und AI Literacy. Ausgehend von der ersten Erwähnung des Begriffs „Digital Literacy“ im Jahr 1997 wird die zunehmende Akademisierung und institutionelle Verankerung dieser Konzepte seit 2015, insbesondere im Bereich Data Literacy, dargestellt. Die Analyse zeigt eine signifikante Zunahme wissenschaftlicher Publikationen zu diesen Themen, was auf eine wachsende gesellschaftliche und bildungspolitische Relevanz hinweist. Dabei werden unterschiedliche Modellansätze wie das Kompetenzrahmenmodell, das Kompetenzstufenmodell sowie spezifische Frameworks wie TPACK, DPACK und AI-PACK vorgestellt, die die Integration von Fachwissen, technologischer Kompetenz und pädagogischer Praxis unterstützen. Besonderes Augenmerk gilt der transdisziplinären und lebenslangen Vermittlung von Kompetenzen, die ethische, kritische und reflektierte Auseinandersetzung mit Daten und Künstlicher Intelligenz einschließen.

@misc{beyer_genki_2024,
	address = {Virtuell},
	title = {{GenKI} im {Lateinunterricht} – {Texte} im {Spracherwerb}},
	url = {https://zenodo.org/records/13902476},
	doi = {10.5281/zenodo.13902476},
	abstract = {Die Fortbildung thematisiert die Integration generativer Künstlicher Intelligenz (genKI) in den Lateinunterricht mit Fokus auf den Spracherwerb. Sie erläutert die Grundlagen von KI und generativer KI, insbesondere deren Fähigkeit, Texte, Bilder, Audios und Videos zu erstellen, zu verändern und zu analysieren. Anhand konkreter Tools wie ChatGPT, Gemini, Claude, Llama oder DALL-E3 werden Einsatzmöglichkeiten für den Unterricht vorgestellt, darunter die Erstellung von didaktisierten Texten, Präsentationen, Übungen und visuellen Materialien. Besonderer Wert wird auf das Prompting und die Prompt Engineering-Strategien gelegt, um qualitativ hochwertige und zielgerichtete Ausgaben zu erzielen. Die Präsentation beleuchtet zudem didaktische Anwendungsszenarien wie Differenzierung, Selbsthilfe für Lernende, Aufgabendesign und das Erstellen von Fehlertexten zur Fehleranalyse. Abschließend werden offene Fragen zu Autorschaft, Leistungserfassung, Bewertungskriterien und KI-Bildung diskutiert, um die reflektierte und verantwortungsvolle Nutzung von genKI im Unterricht zu fördern.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = oct,
	year = {2024},
	keywords = {Latin classes, generative AI, language acquisition, other},
}

Die Fortbildung thematisiert die Integration generativer Künstlicher Intelligenz (genKI) in den Lateinunterricht mit Fokus auf den Spracherwerb. Sie erläutert die Grundlagen von KI und generativer KI, insbesondere deren Fähigkeit, Texte, Bilder, Audios und Videos zu erstellen, zu verändern und zu analysieren. Anhand konkreter Tools wie ChatGPT, Gemini, Claude, Llama oder DALL-E3 werden Einsatzmöglichkeiten für den Unterricht vorgestellt, darunter die Erstellung von didaktisierten Texten, Präsentationen, Übungen und visuellen Materialien. Besonderer Wert wird auf das Prompting und die Prompt Engineering-Strategien gelegt, um qualitativ hochwertige und zielgerichtete Ausgaben zu erzielen. Die Präsentation beleuchtet zudem didaktische Anwendungsszenarien wie Differenzierung, Selbsthilfe für Lernende, Aufgabendesign und das Erstellen von Fehlertexten zur Fehleranalyse. Abschließend werden offene Fragen zu Autorschaft, Leistungserfassung, Bewertungskriterien und KI-Bildung diskutiert, um die reflektierte und verantwortungsvolle Nutzung von genKI im Unterricht zu fördern.

@article{beyer_ki_2024,
	title = {{KI} im altsprachlichen {Unterricht}},
	volume = {5},
	url = {https://www.biejournals.de/index.php/lgnrw/article/view/7309/6512},
	doi = {https://doi.org/10.11576/lgnrw-7309},
	abstract = {Künstliche Intelligenz (KI), insbesondere generative KI (genKI), stellt eine transformative Entwicklung in der digitalen Gesellschaft dar und beeinflusst zunehmend auch den Unterricht in den alten Sprachen. Der Beitrag untersucht die Implikationen von KI für die Klassische Philologie, wobei sowohl die Chancen als auch die Risiken im Hinblick auf Bildung, Ethik, Urheberrecht und Ökologie beleuchtet werden. Generative KI-Systeme wie ChatGPT oder Gemini ermöglichen die automatisierte Übersetzung, Textgenerierung und multimodale Inhaltsproduktion, eröffnen jedoch auch neue Herausforderungen bei der Sicherstellung von Authentizität, Fairness und Nachvollziehbarkeit. Besonders kritisch sind Phänomene wie Halluzinationen, Daten-Bias, Toxizität und die Verbreitung von Deepfakes. Im Kontext des Unterrichts wird die Notwendigkeit einer umfassenden KI-Bildung betont, die von der Grundkenntnis über Funktionsweise und Grenzen von KI bis hin zur kritischen Reflexion und Gestaltung von KI-Anwendungen reicht. Ein Modell für KI-Bildung wird vorgestellt, das sich an den Kompetenzstufen Anfänger, Fortgeschritten, Kompetent und Experte orientiert. Zudem werden Ansätze zur Gestaltung von KI-sensiblen Lehr- und Prüfungsformen diskutiert, die die Eigenleistung stärken und die kritische Auseinandersetzung mit KI fördern. Die Analyse zeigt, dass KI nicht als Ersatz für menschliche Kompetenzen, sondern als Werkzeug zur Reflexion und Erweiterung des Lernprozesses verstanden werden muss, wobei ethische, ökonomische und ökologische Aspekte zentral bleiben.},
	language = {Deutsch},
	number = {1},
	journal = {LGNRW},
	author = {Beyer, Andrea},
	year = {2024},
	keywords = {bold, scientific},
	pages = {9--15},
}

Künstliche Intelligenz (KI), insbesondere generative KI (genKI), stellt eine transformative Entwicklung in der digitalen Gesellschaft dar und beeinflusst zunehmend auch den Unterricht in den alten Sprachen. Der Beitrag untersucht die Implikationen von KI für die Klassische Philologie, wobei sowohl die Chancen als auch die Risiken im Hinblick auf Bildung, Ethik, Urheberrecht und Ökologie beleuchtet werden. Generative KI-Systeme wie ChatGPT oder Gemini ermöglichen die automatisierte Übersetzung, Textgenerierung und multimodale Inhaltsproduktion, eröffnen jedoch auch neue Herausforderungen bei der Sicherstellung von Authentizität, Fairness und Nachvollziehbarkeit. Besonders kritisch sind Phänomene wie Halluzinationen, Daten-Bias, Toxizität und die Verbreitung von Deepfakes. Im Kontext des Unterrichts wird die Notwendigkeit einer umfassenden KI-Bildung betont, die von der Grundkenntnis über Funktionsweise und Grenzen von KI bis hin zur kritischen Reflexion und Gestaltung von KI-Anwendungen reicht. Ein Modell für KI-Bildung wird vorgestellt, das sich an den Kompetenzstufen Anfänger, Fortgeschritten, Kompetent und Experte orientiert. Zudem werden Ansätze zur Gestaltung von KI-sensiblen Lehr- und Prüfungsformen diskutiert, die die Eigenleistung stärken und die kritische Auseinandersetzung mit KI fördern. Die Analyse zeigt, dass KI nicht als Ersatz für menschliche Kompetenzen, sondern als Werkzeug zur Reflexion und Erweiterung des Lernprozesses verstanden werden muss, wobei ethische, ökonomische und ökologische Aspekte zentral bleiben.

@misc{beyer_ki_2024,
	address = {Kiel},
	title = {{KI} im {Spracherwerb} der historischen {Sprache} {Latein}},
	url = {https://zenodo.org/records/14033264},
	doi = {10.5281/zenodo.14033264},
	abstract = {Der Vortrag untersucht die Rolle künstlicher Intelligenz (KI) im Spracherwerb der historischen Sprache Latein unter pädagogischen, psycholinguistischen und didaktischen Aspekten. Er beleuchtet, wie generative KI-Systeme (Large Language Models) zur Unterstützung des textbasierten Spracherwerbs eingesetzt werden können, insbesondere in einem Kontext ohne native Sprecher. Es wird gezeigt, dass KI-Systeme in der Lage sind, Texte zu generieren, zu analysieren, zu paraphrasieren, zu übersetzen und sprachliche Komplexitätsmaße zu ermitteln. Aufbauend auf Modellen des Spracherwerbs moderner Sprachen werden Ansätze zur Modellierung von Erwerbsstufen im Spracherwerb Latein vorgestellt, die auf der Analyse historischer Texte, der Entwicklung fiktiver Sprecher (Personas) und der automatischen Generierung von Lernmaterialien basieren. Zudem wird die Funktion von KI als intelligenter Tutor diskutiert, der individuelles Feedback, adaptive Lernpfade und Learning Analytics ermöglicht. Die Arbeit betont die Notwendigkeit einer fach- und fallspezifischen KI-Bildung, die Kompetenzen in AI Literacy, Data Literacy und Digital Literacy integriert.},
	language = {deu},
	urldate = {2024-11-04},
	author = {Beyer, Andrea},
	month = nov,
	year = {2024},
	keywords = {AI, Historical Language, Language Acquisition, Latin, other},
}

Der Vortrag untersucht die Rolle künstlicher Intelligenz (KI) im Spracherwerb der historischen Sprache Latein unter pädagogischen, psycholinguistischen und didaktischen Aspekten. Er beleuchtet, wie generative KI-Systeme (Large Language Models) zur Unterstützung des textbasierten Spracherwerbs eingesetzt werden können, insbesondere in einem Kontext ohne native Sprecher. Es wird gezeigt, dass KI-Systeme in der Lage sind, Texte zu generieren, zu analysieren, zu paraphrasieren, zu übersetzen und sprachliche Komplexitätsmaße zu ermitteln. Aufbauend auf Modellen des Spracherwerbs moderner Sprachen werden Ansätze zur Modellierung von Erwerbsstufen im Spracherwerb Latein vorgestellt, die auf der Analyse historischer Texte, der Entwicklung fiktiver Sprecher (Personas) und der automatischen Generierung von Lernmaterialien basieren. Zudem wird die Funktion von KI als intelligenter Tutor diskutiert, der individuelles Feedback, adaptive Lernpfade und Learning Analytics ermöglicht. Die Arbeit betont die Notwendigkeit einer fach- und fallspezifischen KI-Bildung, die Kompetenzen in AI Literacy, Data Literacy und Digital Literacy integriert.

@misc{beyer_ki_2024,
	address = {Jena},
	title = {{KI} und {Lateinunterricht}},
	url = {https://zenodo.org/records/10829822},
	doi = {10.5281/zenodo.10829822},
	abstract = {Der Vortrag handelt von der Integration künstlicher Intelligenz (KI), insbesondere generativer KI, in den Lateinunterricht und diskutiert deren pädagogische, ethische und gesellschaftliche Implikationen. Er legt den Fokus auf die Entwicklung von KI-Bildung als zentrale Kompetenz im Rahmen der Digital Literacy. Anhand konkreter Beispielaufgaben wie Übersetzungen, Textgenerierung, Bildanalyse und die Erstellung von Präsentationen wird gezeigt, wie generative KI im Unterricht eingesetzt werden kann, wobei kritische Reflexion über Qualität, Bias, Urheberrecht, Datenschutz und ökologische Auswirkungen notwendig ist. Die Präsentation unterstreicht, dass KI kein Wissen erzeugt, sondern Daten statistisch verarbeitet, und betont die Notwendigkeit von Basiskompetenzen und sinnstiftenden Lernprozessen, um einen souveränen Umgang mit KI-Tools zu ermöglichen. Strategien für KI-sensible Aufgaben, wie die explizite Auseinandersetzung mit Prompts, die Reflexion von Ergebnissen und der Vergleich menschlicher und maschineller Leistungen, werden vorgestellt. Abschließend werden Fortbildungsmöglichkeiten und Ressourcen für Lehrkräfte aufgezeigt, um die fachliche und methodische Kompetenz im Umgang mit KI zu stärken.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = mar,
	year = {2024},
	keywords = {AI Literacy, KI, KI und Prüfen, KI-Bildung, KI-Ethik, Lateinunterricht, other},
}

Der Vortrag handelt von der Integration künstlicher Intelligenz (KI), insbesondere generativer KI, in den Lateinunterricht und diskutiert deren pädagogische, ethische und gesellschaftliche Implikationen. Er legt den Fokus auf die Entwicklung von KI-Bildung als zentrale Kompetenz im Rahmen der Digital Literacy. Anhand konkreter Beispielaufgaben wie Übersetzungen, Textgenerierung, Bildanalyse und die Erstellung von Präsentationen wird gezeigt, wie generative KI im Unterricht eingesetzt werden kann, wobei kritische Reflexion über Qualität, Bias, Urheberrecht, Datenschutz und ökologische Auswirkungen notwendig ist. Die Präsentation unterstreicht, dass KI kein Wissen erzeugt, sondern Daten statistisch verarbeitet, und betont die Notwendigkeit von Basiskompetenzen und sinnstiftenden Lernprozessen, um einen souveränen Umgang mit KI-Tools zu ermöglichen. Strategien für KI-sensible Aufgaben, wie die explizite Auseinandersetzung mit Prompts, die Reflexion von Ergebnissen und der Vergleich menschlicher und maschineller Leistungen, werden vorgestellt. Abschließend werden Fortbildungsmöglichkeiten und Ressourcen für Lehrkräfte aufgezeigt, um die fachliche und methodische Kompetenz im Umgang mit KI zu stärken.

@misc{beyer_ki_2024,
	address = {Virtuell},
	title = {{KI} und Übersetzen – {Literaturunterricht} {Latein}},
	url = {https://zenodo.org/records/14063591},
	doi = {10.5281/zenodo.14063591},
	abstract = {Diese Fortbildung geht auf die Rolle von Künstlicher Intelligenz (KI) im Lateinunterricht, insbesondere im Kontext des Übersetzens literarischer Texte, ein. Ausgehend von verschiedenen Übersetzungsmodellen – linguistisch, handlungstheoretisch und systemisch-kultursemiotisch – werden die Grundlagen für eine kritische Beurteilung von Übersetzungen erarbeitet. Dabei wird die Komplexität des Übersetzungsprozesses hervorgehoben, der nicht nur sprachliche, sondern auch kulturelle, stilistische und pragmatische Aspekte umfasst. Der Beitrag beleuchtet die Entwicklung maschineller Übersetzung, von regelbasierten und statistischen Ansätzen hin zu neuronalen Netzwerken und generativen Sprachmodellen, und analysiert deren Stärken und Grenzen, insbesondere im Hinblick auf literarische Texte. Besonderes Augenmerk gilt der kritischen Nutzung von KI-Übersetzungshilfen im Unterricht, die als unterstützende Werkzeuge für die Textanalyse, Wortschatzarbeit, Syntax- und Rhythmusanalyse sowie Kontexterklärungen dienen können. Die Diskussion zeigt, dass KI-Übersetzungen nicht als Ersatz für menschliche Übersetzungen gelten dürfen, sondern vielmehr als Instrumente zur Förderung von Translationskompetenz und kritischem Denken im Lernprozess genutzt werden sollten.},
	language = {deu},
	urldate = {2024-11-12},
	author = {Beyer, Andrea},
	month = nov,
	year = {2024},
	keywords = {AI, Latin, Machine Translation, Translation, other},
}

Diese Fortbildung geht auf die Rolle von Künstlicher Intelligenz (KI) im Lateinunterricht, insbesondere im Kontext des Übersetzens literarischer Texte, ein. Ausgehend von verschiedenen Übersetzungsmodellen – linguistisch, handlungstheoretisch und systemisch-kultursemiotisch – werden die Grundlagen für eine kritische Beurteilung von Übersetzungen erarbeitet. Dabei wird die Komplexität des Übersetzungsprozesses hervorgehoben, der nicht nur sprachliche, sondern auch kulturelle, stilistische und pragmatische Aspekte umfasst. Der Beitrag beleuchtet die Entwicklung maschineller Übersetzung, von regelbasierten und statistischen Ansätzen hin zu neuronalen Netzwerken und generativen Sprachmodellen, und analysiert deren Stärken und Grenzen, insbesondere im Hinblick auf literarische Texte. Besonderes Augenmerk gilt der kritischen Nutzung von KI-Übersetzungshilfen im Unterricht, die als unterstützende Werkzeuge für die Textanalyse, Wortschatzarbeit, Syntax- und Rhythmusanalyse sowie Kontexterklärungen dienen können. Die Diskussion zeigt, dass KI-Übersetzungen nicht als Ersatz für menschliche Übersetzungen gelten dürfen, sondern vielmehr als Instrumente zur Förderung von Translationskompetenz und kritischem Denken im Lernprozess genutzt werden sollten.

@misc{beyer_nlp-infrastruktur_2024,
	address = {Berlin},
	title = {Eine {NLP}-{Infrastruktur} für {KI}-skeptische {User}},
	url = {https://zenodo.org/records/12199989},
	doi = {10.5281/zenodo.12199989},
	abstract = {Das interdisziplinäre DFG-Projekt Daidalos entwickelt eine webbasierte NLP-Infrastruktur, die Forschenden aus der Klassischen Philologie und angrenzenden Disziplinen ermöglicht, verschiedene NLP-Methoden auf selbst zusammengestellte, kleine, asynchrone literarische Korpora im Kontext sprach- und literaturwissenschaftlicher Forschungsvorhaben anzuwenden. Ziel ist es, durch userzentrierte Gestaltung einer grafischen Benutzeroberfläche, kuratierte Workflows in Jupyter Notebooks, eine Literatur- und Tooldatenbank sowie Community-Workshops und Beratung die digitale Forschungskompetenz der User zu stärken. Besonderes Augenmerk liegt auf Transparenz und Nachnutzbarkeit durch die Implementierung von Model Cards und Data Sheets sowie einer dokumentierten Evaluation der eingesetzten Methoden. Der Ansatz wird durch Forschungstandems und interdisziplinäre Kooperationen unterstützt, um die Herausforderungen von close und distant reading, geringer digitaler Kompetenz und fehlenden Lehr- und Lernmaterialien zu bewältigen.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea and Kotschka, Florian},
	month = jun,
	year = {2024},
	keywords = {Classics, Computer and information sciences, Literary Studies, Literature studies, NLP, other},
}

Das interdisziplinäre DFG-Projekt Daidalos entwickelt eine webbasierte NLP-Infrastruktur, die Forschenden aus der Klassischen Philologie und angrenzenden Disziplinen ermöglicht, verschiedene NLP-Methoden auf selbst zusammengestellte, kleine, asynchrone literarische Korpora im Kontext sprach- und literaturwissenschaftlicher Forschungsvorhaben anzuwenden. Ziel ist es, durch userzentrierte Gestaltung einer grafischen Benutzeroberfläche, kuratierte Workflows in Jupyter Notebooks, eine Literatur- und Tooldatenbank sowie Community-Workshops und Beratung die digitale Forschungskompetenz der User zu stärken. Besonderes Augenmerk liegt auf Transparenz und Nachnutzbarkeit durch die Implementierung von Model Cards und Data Sheets sowie einer dokumentierten Evaluation der eingesetzten Methoden. Der Ansatz wird durch Forschungstandems und interdisziplinäre Kooperationen unterstützt, um die Herausforderungen von close und distant reading, geringer digitaler Kompetenz und fehlenden Lehr- und Lernmaterialien zu bewältigen.

@inproceedings{beyer_daidalos_2024,
	address = {Passau},
	title = {Daidalos: {Wie} viel {Methodenkompetenz} braucht ein {User}?},
	url = {https://zenodo.org/records/10698299},
	doi = {10.5281/zenodo.10698299},
	abstract = {Das Daidalos-Projekt zielt darauf ab, eine Forschungsinfrastruktur für die deutschsprachige Klassische Philologie zu entwickeln, die den Übergang von analogen zu digitalen Forschungsmethoden unterstützt. Es adressiert die bestehende Kompetenzlücke im Bereich der Digital Humanities (DH), insbesondere im Hinblick auf Natural Language Processing (NLP), indem es Forschenden mit unterschiedlichen Voraussetzungen die Möglichkeit bietet, selbstständig Forschungsfragen zu entwickeln und DH-Methoden sinnvoll einzusetzen. Daidalos integriert bestehende Tools wie LatinCy und bietet über ein auf forschendes Lernen basierendes Lernangebot mit drei Niveaustufen – von strukturierten Demo-Workflows bis hin zu freier Methodenkonfiguration. Zentrales Element ist die Bildung von Forschungstandems aus erfahrenen und weniger erfahrenen Nutzern, die gemeinsam Forschungsfragen bearbeiten, Ergebnisse publizieren und so als authentische Lernmaterialien dienen. Durch die Entwicklung einer fachspezifischen Digital Literacy in einer Community of Practice soll die Akzeptanz der Infrastruktur in der Fachgemeinschaft gestärkt und die Integration von DH-Methoden in Forschung und Lehre nachhaltig gefördert werden.},
	booktitle = {Book of {Abstracts} - {DHd2024}},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = feb,
	year = {2024},
	keywords = {scientific},
	pages = {336--338},
}

Das Daidalos-Projekt zielt darauf ab, eine Forschungsinfrastruktur für die deutschsprachige Klassische Philologie zu entwickeln, die den Übergang von analogen zu digitalen Forschungsmethoden unterstützt. Es adressiert die bestehende Kompetenzlücke im Bereich der Digital Humanities (DH), insbesondere im Hinblick auf Natural Language Processing (NLP), indem es Forschenden mit unterschiedlichen Voraussetzungen die Möglichkeit bietet, selbstständig Forschungsfragen zu entwickeln und DH-Methoden sinnvoll einzusetzen. Daidalos integriert bestehende Tools wie LatinCy und bietet über ein auf forschendes Lernen basierendes Lernangebot mit drei Niveaustufen – von strukturierten Demo-Workflows bis hin zu freier Methodenkonfiguration. Zentrales Element ist die Bildung von Forschungstandems aus erfahrenen und weniger erfahrenen Nutzern, die gemeinsam Forschungsfragen bearbeiten, Ergebnisse publizieren und so als authentische Lernmaterialien dienen. Durch die Entwicklung einer fachspezifischen Digital Literacy in einer Community of Practice soll die Akzeptanz der Infrastruktur in der Fachgemeinschaft gestärkt und die Integration von DH-Methoden in Forschung und Lehre nachhaltig gefördert werden.

@misc{beyer_digitale_2024,
	address = {Eichstätt},
	title = {Digitale {Methoden} in der {Klassischen} {Philologie}},
	url = {https://zenodo.org/records/10529746},
	doi = {10.5281/zenodo.10529746},
	abstract = {Der Vortrag im Colloquium bietet eine Einführung in digitale Methoden der Klassischen Philologie mit dem Schwerpunkt auf Natural Language Processing (NLP) und deren Anwendung in der Literaturwissenschaft. Er beschreibt die grundlegenden Schritte der digitalen Textverarbeitung, beginnend mit der Erfassung und Vorverarbeitung von Texten über die Annotation und den Aufbau von Korpora bis hin zur Durchführung erster Korpusanalysen. Anhand von vier Fallstudien wird die Anwendung von NLP-Verfahren in der Forschung demonstriert: die Autorschaftsanalyse des Commentariolum petitionis, die maschinelle Erkennung von intertextuellen Anspielungen zwischen lateinischen und griechischen Texten, die Wortfeldanalyse zum Konzept „Schmerz“ in der klassischen griechischen Literatur sowie die Identifikation philosophischer Textpassagen ohne explizite Zitate. Die vorgestellten Methoden umfassen Word Embeddings, Topic Modelling, Multilingual Sentence Embeddings und Visualisierungstechniken. Das Dokument diskutiert zudem den aktuellen Stand und die Herausforderungen der Digital Humanities in der deutschsprachigen Klassischen Philologie, betont die Bedeutung fachübergreifender Kompetenzen und kollaborativer Forschungskulturen sowie die zunehmende Förderungswürdigkeit methodischer Vielfalt. Abschließend wird auf das Daidalos-Projekt als Beispiel für zukunftsorientierte Forschungskooperation hingewiesen.},
	language = {deu},
	urldate = {2024-03-18},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jan,
	year = {2024},
	keywords = {Classics, computational literary studies, digital humanities, natural language processing, other},
}

Der Vortrag im Colloquium bietet eine Einführung in digitale Methoden der Klassischen Philologie mit dem Schwerpunkt auf Natural Language Processing (NLP) und deren Anwendung in der Literaturwissenschaft. Er beschreibt die grundlegenden Schritte der digitalen Textverarbeitung, beginnend mit der Erfassung und Vorverarbeitung von Texten über die Annotation und den Aufbau von Korpora bis hin zur Durchführung erster Korpusanalysen. Anhand von vier Fallstudien wird die Anwendung von NLP-Verfahren in der Forschung demonstriert: die Autorschaftsanalyse des Commentariolum petitionis, die maschinelle Erkennung von intertextuellen Anspielungen zwischen lateinischen und griechischen Texten, die Wortfeldanalyse zum Konzept „Schmerz“ in der klassischen griechischen Literatur sowie die Identifikation philosophischer Textpassagen ohne explizite Zitate. Die vorgestellten Methoden umfassen Word Embeddings, Topic Modelling, Multilingual Sentence Embeddings und Visualisierungstechniken. Das Dokument diskutiert zudem den aktuellen Stand und die Herausforderungen der Digital Humanities in der deutschsprachigen Klassischen Philologie, betont die Bedeutung fachübergreifender Kompetenzen und kollaborativer Forschungskulturen sowie die zunehmende Förderungswürdigkeit methodischer Vielfalt. Abschließend wird auf das Daidalos-Projekt als Beispiel für zukunftsorientierte Forschungskooperation hingewiesen.

@misc{beyer_fach-_2024,
	address = {Berlin},
	type = {Zenodo},
	title = {Fach- und fallspezifische {KI}-{Bildung} in den {Geisteswissenschaften}},
	url = {https://zenodo.org/records/13757021},
	doi = {10.5281/zenodo.13757021},
	abstract = {Der Vortrag stellt ein Konzept zur fach- und fallspezifischen KI-Bildung in den Geisteswissenschaften vor, das darauf abzielt, Studierende, Lehrende und Forschende mit den notwendigen Kompetenzen auszustatten, um Künstliche Intelligenz kritisch zu reflektieren, verantwortungsvoll zu nutzen und in der eigenen Disziplin anzuwenden. Ausgehend von der Herausforderung, dass KI-Technologien wie Deepfakes, Bias und algorithmische Entscheidungen gesellschaftliche und wissenschaftliche Praxis beeinflussen, wird ein gestuftes Kompetenzmodell vorgestellt, das auf den Dimensionen Komplexität der Aufgabe, Autonomiegrad und kognitivem Bereich aufbaut. Anhand eines Fallbeispiels aus der Klassischen Philologie – der Analyse ausgelassener historischer Ereignisse in antiken Texten – wird die Integration von AI-Literacy (z. B. Named Entity Recognition), Data-Literacy (z. B. Annotationen, Datenformate) und Digital-Literacy (z. B. Datei- und Ordnerstrukturen) veranschaulicht. Das Konzept betont die Notwendigkeit einer individualisierten, anwendungsbezogenen und reflektierenden Bildung, die über reine Technikkompetenz hinausgeht und auf die spezifischen Anforderungen der Fachdisziplinen eingeht. Die praktische Umsetzung wird durch strukturierte Materialien, Austauschformate wie Tandems und Workshops sowie die Einbindung von Forschungsprojekten unterstrichen. Das Fazit unterstreicht, dass das erforderliche Kompetenzniveau sich aus dem Forschungs- oder Lernziel ergibt und eine flexible, nicht linear aufgebaute Stufung erfordert, die auf dem Prinzip „know your data!“ basiert.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2024},
	keywords = {Artificial intelligence, Classics, Educational sciences, Languages and literature, Natural language processing, bold, other},
}

Der Vortrag stellt ein Konzept zur fach- und fallspezifischen KI-Bildung in den Geisteswissenschaften vor, das darauf abzielt, Studierende, Lehrende und Forschende mit den notwendigen Kompetenzen auszustatten, um Künstliche Intelligenz kritisch zu reflektieren, verantwortungsvoll zu nutzen und in der eigenen Disziplin anzuwenden. Ausgehend von der Herausforderung, dass KI-Technologien wie Deepfakes, Bias und algorithmische Entscheidungen gesellschaftliche und wissenschaftliche Praxis beeinflussen, wird ein gestuftes Kompetenzmodell vorgestellt, das auf den Dimensionen Komplexität der Aufgabe, Autonomiegrad und kognitivem Bereich aufbaut. Anhand eines Fallbeispiels aus der Klassischen Philologie – der Analyse ausgelassener historischer Ereignisse in antiken Texten – wird die Integration von AI-Literacy (z. B. Named Entity Recognition), Data-Literacy (z. B. Annotationen, Datenformate) und Digital-Literacy (z. B. Datei- und Ordnerstrukturen) veranschaulicht. Das Konzept betont die Notwendigkeit einer individualisierten, anwendungsbezogenen und reflektierenden Bildung, die über reine Technikkompetenz hinausgeht und auf die spezifischen Anforderungen der Fachdisziplinen eingeht. Die praktische Umsetzung wird durch strukturierte Materialien, Austauschformate wie Tandems und Workshops sowie die Einbindung von Forschungsprojekten unterstrichen. Das Fazit unterstreicht, dass das erforderliche Kompetenzniveau sich aus dem Forschungs- oder Lernziel ergibt und eine flexible, nicht linear aufgebaute Stufung erfordert, die auf dem Prinzip „know your data!“ basiert.

@misc{beyer_generative_2024,
	address = {Virtuell},
	title = {Generative {KI} und ihre {Bedeutung} für {Bewertungskontexte}},
	url = {https://zenodo.org/records/10569149},
	doi = {10.5281/zenodo.10569149},
	abstract = {Generative Künstliche Intelligenz (KI) stellt eine bedeutende Herausforderung und Chance für den schulischen Bewertungskontext dar. Der Vortrag analysiert die Auswirkungen von generativer KI auf Lehr- und Lernprozesse, insbesondere im Hinblick auf die Leistungsmessung und -bewertung im KI-Zeitalter. Er beleuchtet die Funktionsweise und Anwendungsfelder von Large Language Models (LLM), deren Stärken wie schnelle Textgenerierung sowie gravierende Schwächen wie Bias, Halluzinationen und Transparenzprobleme. Im Kontext der Bildung wird KI-Bildung als zentrale Dimension der Digitalkompetenz vorgestellt, die Kompetenzen im Verstehen, Anwenden, Bewerten und Gestalten von KI umfasst. Die Arbeit diskutiert ethische, ökologische und rechtliche Aspekte wie Urheberrecht, Datenschutz und ökologische Belastungen durch die KI-Entwicklung. Zudem werden konkrete Beispielaufgaben analysiert, um die Auswirkungen auf die Gestaltung von Prüfungen und die Bewertung von Lernleistungen zu verdeutlichen.},
	language = {deu},
	urldate = {2024-03-18},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jan,
	year = {2024},
	keywords = {AI, AI Literacy, KI Ethik, KI und Prüfungen, KI-Bildung, generative KI, other},
}

Generative Künstliche Intelligenz (KI) stellt eine bedeutende Herausforderung und Chance für den schulischen Bewertungskontext dar. Der Vortrag analysiert die Auswirkungen von generativer KI auf Lehr- und Lernprozesse, insbesondere im Hinblick auf die Leistungsmessung und -bewertung im KI-Zeitalter. Er beleuchtet die Funktionsweise und Anwendungsfelder von Large Language Models (LLM), deren Stärken wie schnelle Textgenerierung sowie gravierende Schwächen wie Bias, Halluzinationen und Transparenzprobleme. Im Kontext der Bildung wird KI-Bildung als zentrale Dimension der Digitalkompetenz vorgestellt, die Kompetenzen im Verstehen, Anwenden, Bewerten und Gestalten von KI umfasst. Die Arbeit diskutiert ethische, ökologische und rechtliche Aspekte wie Urheberrecht, Datenschutz und ökologische Belastungen durch die KI-Entwicklung. Zudem werden konkrete Beispielaufgaben analysiert, um die Auswirkungen auf die Gestaltung von Prüfungen und die Bewertung von Lernleistungen zu verdeutlichen.

@misc{beyer_nlp-infrastruktur_2024,
	address = {Berlin},
	title = {{NLP}-{Infrastruktur} für die {Klassische} {Philologie}},
	url = {https://zenodo.org/records/10474686},
	doi = {10.5281/zenodo.10474686},
	abstract = {Das Daidalos-Projekt zielt auf den Aufbau einer interaktiven Forschungs- und Lerninfrastruktur für die Klassische Philologie, die den Einsatz von Natural Language Processing (NLP) für lateinische und griechische literarische Texte ermöglicht. Ausgehend von dem Bedarf nach einer fachspezifischen Digital Literacy und der Notwendigkeit, NLP-Methoden ohne tiefgehende Programmierkenntnisse zugänglich zu machen, wird eine No-Code-Plattform entwickelt, die Forschende dabei unterstützt, beliebig große Textkorpora zusammenzustellen, zu analysieren, zu visualisieren und zu teilen. Die Infrastruktur integriert etablierte NLP-Techniken wie Named Entity Recognition, Lemmatisierung und Word Embeddings. Durch die Zusammenarbeit mit Forschungstandems aus erfahrenen und weniger erfahrenen Nutzern wird die Plattform an authentischen Forschungsfragen ausgerichtet und kontinuierlich an den Bedürfnissen der Community weiterentwickelt. Zentrale Aspekte sind die Förderung situierten Lernens, die Entwicklung einer Community of Practice, die Evaluation anhand von Testdatensätzen und Nutzungsstudien sowie die Veröffentlichung von Ergebnissen in offener Form.},
	language = {deu},
	urldate = {2024-03-18},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jan,
	year = {2024},
	keywords = {Classics, computational literary studies, natural language processing, other, research infrastructure},
}

Das Daidalos-Projekt zielt auf den Aufbau einer interaktiven Forschungs- und Lerninfrastruktur für die Klassische Philologie, die den Einsatz von Natural Language Processing (NLP) für lateinische und griechische literarische Texte ermöglicht. Ausgehend von dem Bedarf nach einer fachspezifischen Digital Literacy und der Notwendigkeit, NLP-Methoden ohne tiefgehende Programmierkenntnisse zugänglich zu machen, wird eine No-Code-Plattform entwickelt, die Forschende dabei unterstützt, beliebig große Textkorpora zusammenzustellen, zu analysieren, zu visualisieren und zu teilen. Die Infrastruktur integriert etablierte NLP-Techniken wie Named Entity Recognition, Lemmatisierung und Word Embeddings. Durch die Zusammenarbeit mit Forschungstandems aus erfahrenen und weniger erfahrenen Nutzern wird die Plattform an authentischen Forschungsfragen ausgerichtet und kontinuierlich an den Bedürfnissen der Community weiterentwickelt. Zentrale Aspekte sind die Förderung situierten Lernens, die Entwicklung einer Community of Practice, die Evaluation anhand von Testdatensätzen und Nutzungsstudien sowie die Veröffentlichung von Ergebnissen in offener Form.

@misc{beyer_nlp-methoden_2024,
	address = {Berlin},
	title = {{NLP}-{Methoden} in der {Klassischen} {Philologie}: {Word} {Embeddings}},
	shorttitle = {{NLP}-{Methoden} in der {Klassischen} {Philologie}},
	url = {https://zenodo.org/records/11582358},
	doi = {10.5281/zenodo.11582358},
	abstract = {Der Vortrag präsentiert den Einsatz von Word Embeddings als Methode des Natural Language Processing (NLP) in der Klassischen Philologie. Er erläutert die Grundlagen der Vektorrepräsentation von Wörtern im Kontext digitaler Textanalyse und zeigt anhand konkreter Fallstudien deren Anwendung. Die vorgestellten Methoden, darunter Textklassifikation, Lemmatisierung, Part-of-Speech-Tagging und Topic Modelling, ermöglichen eine quantitative und datenbasierte Untersuchung altsprachlicher Texte. Die Ergebnisse verdeutlichen, wie vektorbasierte Modelle neue Perspektiven auf traditionelle philologische Fragen eröffnen, gleichzeitig aber auch die Herausforderungen durch Datenbias, morphologische Komplexität und die Beschränktheit bestimmter linguistischer Analogien mit sich bringen.},
	language = {deu},
	urldate = {2024-06-12},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jun,
	year = {2024},
	keywords = {Artificial intelligence, Classics, Computer and information sciences, Languages and literature, Linguistics, Representation Learning, Word Embeddings, other},
}

Der Vortrag präsentiert den Einsatz von Word Embeddings als Methode des Natural Language Processing (NLP) in der Klassischen Philologie. Er erläutert die Grundlagen der Vektorrepräsentation von Wörtern im Kontext digitaler Textanalyse und zeigt anhand konkreter Fallstudien deren Anwendung. Die vorgestellten Methoden, darunter Textklassifikation, Lemmatisierung, Part-of-Speech-Tagging und Topic Modelling, ermöglichen eine quantitative und datenbasierte Untersuchung altsprachlicher Texte. Die Ergebnisse verdeutlichen, wie vektorbasierte Modelle neue Perspektiven auf traditionelle philologische Fragen eröffnen, gleichzeitig aber auch die Herausforderungen durch Datenbias, morphologische Komplexität und die Beschränktheit bestimmter linguistischer Analogien mit sich bringen.

@misc{beyer_reflexion_2024,
	address = {Wuppertal},
	title = {Reflexion  mit und über {KI} im {AU}},
	url = {https://zenodo.org/records/10909593},
	doi = {10.5281/zenodo.10909593},
	abstract = {Der Vortrag behandelt die Reflexion mit und über Künstliche Intelligenz (KI) im Kontext des altsprachlichen Unterrichts. Er bietet eine Einführung in KI und generative KI, erläutert deren technische Grundlagen, Anwendungsfelder und ethische sowie gesellschaftliche Implikationen wie Bias, Desinformation, Urheberrechtsfragen und ökologische Belastungen. Im Zentrum steht die Entwicklung von KI-Bildung als zentrale Kompetenzdimension innerhalb der Digital Literacy, die spezifisch auf die alten Sprachen ausgerichtet ist. Dazu werden Kompetenzstufenmodelle vorgestellt, die von der Grundkenntnis über Anwendung bis hin zur kritischen Bewertung und Gestaltung von KI-Systemen reichen. Praktische Beispielaufgaben aus den Bereichen Sprach-, Literatur-, Kultur- und Translationskompetenz verdeutlichen den integrativen Einsatz von KI im Unterricht. Es wird gezeigt, dass generative KI kein eigenständiges Wissen erzeugt, sondern auf Daten basiert. Dies unterstreicht die Notwendigkeit einer kritischen Auseinandersetzung mit den Ergebnissen. Abschließend werden Ansätze für einen kI-sensiblen Unterricht vorgestellt, der die menschliche Urteilskraft, kognitive Kompetenzen und sinnstiftende Lernprozesse in den Mittelpunkt stellt.},
	language = {deu},
	urldate = {2024-04-02},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = apr,
	year = {2024},
	keywords = {AI Literacy, Artificial intelligence, Classics, Second Language Acquisition, other},
}

Der Vortrag behandelt die Reflexion mit und über Künstliche Intelligenz (KI) im Kontext des altsprachlichen Unterrichts. Er bietet eine Einführung in KI und generative KI, erläutert deren technische Grundlagen, Anwendungsfelder und ethische sowie gesellschaftliche Implikationen wie Bias, Desinformation, Urheberrechtsfragen und ökologische Belastungen. Im Zentrum steht die Entwicklung von KI-Bildung als zentrale Kompetenzdimension innerhalb der Digital Literacy, die spezifisch auf die alten Sprachen ausgerichtet ist. Dazu werden Kompetenzstufenmodelle vorgestellt, die von der Grundkenntnis über Anwendung bis hin zur kritischen Bewertung und Gestaltung von KI-Systemen reichen. Praktische Beispielaufgaben aus den Bereichen Sprach-, Literatur-, Kultur- und Translationskompetenz verdeutlichen den integrativen Einsatz von KI im Unterricht. Es wird gezeigt, dass generative KI kein eigenständiges Wissen erzeugt, sondern auf Daten basiert. Dies unterstreicht die Notwendigkeit einer kritischen Auseinandersetzung mit den Ergebnissen. Abschließend werden Ansätze für einen kI-sensiblen Unterricht vorgestellt, der die menschliche Urteilskraft, kognitive Kompetenzen und sinnstiftende Lernprozesse in den Mittelpunkt stellt.

@misc{schulz_seflag_2024,
	title = {{SEFLAG}. {Systematic} {Evaluation} {Framework} for {NLP} {Models} and {Datasets} in {Latin} and {Ancient} {Greek}},
	url = {https://zenodo.org/records/14012948},
	doi = {10.5281/zenodo.14012948},
	abstract = {Das SEFLAG-Framework, präsentiert von Konstantin Schulz und Florian Deichsler (Humboldt-Universität zu Berlin), stellt ein systematisches Evaluationsframework für NLP-Modelle und -Datensätze für Latein und Altgriechisch vor. Die Hauptziele sind die Bewertung und Dokumentation bestehender NLP-Ressourcen, die Auswahl passender Modelle für Forschungsvorhaben sowie die Förderung der Interoperabilität durch standardisierte Annotationsrichtlinien. Diese Initiative reagiert auf den steigenden Einsatz von NLP in der Literaturwissenschaft historischer Sprachen und die damit verbundenen Herausforderungen, wie die Diversität an Modellen, die fehlende zentrale Verwaltung von Ressourcen und die Notwendigkeit einer systematischen Evaluation.

SEFLAG konzentriert sich zunächst auf Lemmatisierung und Named Entity Recognition (NER) und plant zukünftig die Erweiterung um weitere NLP-Tasks. Die Ergebnisse werden auf Plattformen wie Hugging Face veröffentlicht und zielen darauf ab, der Forschungsgemeinschaft Zeit und Ressourcen zu sparen. Durch Dokumentation und Benchmarking wird eine nachhaltige Infrastruktur geschaffen, die Forschenden in den Bereichen Philologie, Geschichte und Archäologie zugutekommt und Innovationen im Bereich der historischen NLP-Methoden fördert.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Schulz, Konstantin and Deichsler, Florian},
	month = oct,
	year = {2024},
	keywords = {Artificial intelligence, Computer and information sciences, Languages and literature, Linguistics, Natural language processing},
}

Das SEFLAG-Framework, präsentiert von Konstantin Schulz und Florian Deichsler (Humboldt-Universität zu Berlin), stellt ein systematisches Evaluationsframework für NLP-Modelle und -Datensätze für Latein und Altgriechisch vor. Die Hauptziele sind die Bewertung und Dokumentation bestehender NLP-Ressourcen, die Auswahl passender Modelle für Forschungsvorhaben sowie die Förderung der Interoperabilität durch standardisierte Annotationsrichtlinien. Diese Initiative reagiert auf den steigenden Einsatz von NLP in der Literaturwissenschaft historischer Sprachen und die damit verbundenen Herausforderungen, wie die Diversität an Modellen, die fehlende zentrale Verwaltung von Ressourcen und die Notwendigkeit einer systematischen Evaluation. SEFLAG konzentriert sich zunächst auf Lemmatisierung und Named Entity Recognition (NER) und plant zukünftig die Erweiterung um weitere NLP-Tasks. Die Ergebnisse werden auf Plattformen wie Hugging Face veröffentlicht und zielen darauf ab, der Forschungsgemeinschaft Zeit und Ressourcen zu sparen. Durch Dokumentation und Benchmarking wird eine nachhaltige Infrastruktur geschaffen, die Forschenden in den Bereichen Philologie, Geschichte und Archäologie zugutekommt und Innovationen im Bereich der historischen NLP-Methoden fördert.

@book{unesco_ai_2024,
	title = {{AI} competency framework for teachers},
	isbn = {978-92-3-100707-1},
	url = {https://unesdoc.unesco.org/ark:/48223/pf0000391104?posInSet=1&queryId=310ee262-446b-4d16-88a8-b6bfd4f77887},
	urldate = {2024-09-11},
	publisher = {UNESDOC},
	author = {Miao, Fengchun and Cukurova, Mutlu},
	editor = {{UNESCO}},
	year = {2024},
	doi = {https://doi.org/10.54675/ZJTE2084},
}

@incollection{la_veglia_being_2024,
	address = {Berlin, Boston},
	title = {Being a {Classicist} in the {Digital} {Age}},
	url = {https://library.oapen.org/bitstream/handle/20.500.12657/98796/1/9783111070162.pdf},
	doi = {10.1515/9783111070162},
	language = {en},
	booktitle = {Digital {Papyrology} {III}: {The} {Digital} {Critical} {Edition} of {Greek} {Papyri}: {Issues}, {Projects}, and {Perspectives}},
	publisher = {De Gruyter},
	author = {La Veglia, Andrea},
	editor = {Reggiani, Nicola},
	year = {2024},
	pages = {49--70},
}

@misc{schulz_einfuhrung_2024,
	address = {Berlin},
	type = {Zenodo},
	title = {Einführung in {Natural} {Language} {Processing} anhand von {Plinius}' {Brief} 1,8},
	url = {https://zenodo.org/records/13907150},
	doi = {10.5281/zenodo.13907150},
	abstract = {Die Präsentation behandelt die methodische Analyse von Plinius' Briefen, insbesondere die Herausforderungen der Selbstdarstellung in epist. 1,8. Verschiedene linguistische und computerlinguistische Verfahren werden eingesetzt, um den Text digital zu untersuchen. Zunächst wird die Lemmatisierung eingeführt, die die Reduzierung flektierter Formen auf ihre Grundform beschreibt. Dadurch wird eine bessere Durchsuchbarkeit und Vergleichbarkeit von Texten ermöglicht. Darauf folgt das Part-of-Speech-Tagging, bei dem Wortarten im Text annotiert werden. Probleme wie die Mehrdeutigkeit und Schwierigkeiten in den Trainingsdaten des Algorithmus werden hervorgehoben. Zudem wird eine Sentimentanalyse durchgeführt, die emotionale Bewertungen und Meinungen im Text identifiziert. Ferner kommen Word Embeddings zum Einsatz, um semantische Beziehungen zwischen Wörtern darzustellen. Abschließend wird die Syntax des Plinius-Briefs mittels Treebanking, also der systematischen Erfassung syntaktischer Abhängigkeiten zwischen Wörtern in einem Satz, analysiert. Der Vortrag zeigt, wie digitale Werkzeuge klassisch-philologische Fragestellungen unterstützen und neue Erkenntnisse über Plinius' Werk ermöglichen, insbesondere in Bezug auf die Darstellung von Ruhm, Bescheidenheit und Selbstreflexion.},
	language = {deu},
	urldate = {2024-10-09},
	author = {Schulz, Konstantin},
	month = may,
	year = {2024},
	keywords = {Classics, Computational Linguistics, Languages and literature, Latin, Natural Language Processing, other},
}

Die Präsentation behandelt die methodische Analyse von Plinius' Briefen, insbesondere die Herausforderungen der Selbstdarstellung in epist. 1,8. Verschiedene linguistische und computerlinguistische Verfahren werden eingesetzt, um den Text digital zu untersuchen. Zunächst wird die Lemmatisierung eingeführt, die die Reduzierung flektierter Formen auf ihre Grundform beschreibt. Dadurch wird eine bessere Durchsuchbarkeit und Vergleichbarkeit von Texten ermöglicht. Darauf folgt das Part-of-Speech-Tagging, bei dem Wortarten im Text annotiert werden. Probleme wie die Mehrdeutigkeit und Schwierigkeiten in den Trainingsdaten des Algorithmus werden hervorgehoben. Zudem wird eine Sentimentanalyse durchgeführt, die emotionale Bewertungen und Meinungen im Text identifiziert. Ferner kommen Word Embeddings zum Einsatz, um semantische Beziehungen zwischen Wörtern darzustellen. Abschließend wird die Syntax des Plinius-Briefs mittels Treebanking, also der systematischen Erfassung syntaktischer Abhängigkeiten zwischen Wörtern in einem Satz, analysiert. Der Vortrag zeigt, wie digitale Werkzeuge klassisch-philologische Fragestellungen unterstützen und neue Erkenntnisse über Plinius' Werk ermöglichen, insbesondere in Bezug auf die Darstellung von Ruhm, Bescheidenheit und Selbstreflexion.

@misc{faltin_automatisierte_2024,
	address = {Kiel},
	type = {Zenodo},
	title = {Automatisierte {Kategorisierung} mittellateinischer {Bittbriefe} an den {Papst}},
	url = {https://zenodo.org/records/13628818},
	doi = {10.5281/zenodo.13628818},
	abstract = {Die Präsentation ist im Rahmen der 27. Aquilonia an der Christian-Albrechts-Universität zu Kiel entstanden.},
	language = {Deutsch},
	urldate = {2024-09-04},
	author = {Faltin, Nico and Schulz, Konstantin},
	month = jul,
	year = {2024},
	keywords = {bold, other},
}

Die Präsentation ist im Rahmen der 27. Aquilonia an der Christian-Albrechts-Universität zu Kiel entstanden.

@misc{dedema_collective_2024,
	title = {The collective use and perceptions of generative {AI} tools in digital humanities research: {Survey}-based results},
	url = {https://arxiv.org/pdf/2404.12458},
	publisher = {arXiv},
	author = {Dedema, Meredith and Ma, Rongqian},
	year = {2024},
}

@inproceedings{penedo_fineweb_2024,
	title = {The fineweb datasets: {Decanting} the web for the finest text data at scale},
	volume = {37},
	url = {https://proceedings.neurips.cc/paper_files/paper/2024/file/370df50ccfdf8bde18f8f9c2d9151bda-Paper-Datasets_and_Benchmarks_Track.pdf},
	doi = {10.52202/079017-0970},
	booktitle = {Advances in {Neural} {Information} {Processing} {Systems}},
	author = {Penedo, Guilherme and Kydlíček, Hynek and Lozhkov, Anton and Mitchell, Margaret and Raffel, Colin A and Von Werra, Leandro and Wolf, Thomas},
	year = {2024},
	pages = {30811--30849},
}

@article{stopponi_agree_2024,
	title = {{AGREE}: a new benchmark for the evaluation of distributional semantic models of ancient {Greek}},
	shorttitle = {{AGREE}},
	url = {https://research.rug.nl/en/publications/agree-a-new-benchmark-for-the-evaluation-of-distributional-semant},
	doi = {10.1093/llc/fqad087},
	urldate = {2024-04-05},
	journal = {Digital Scholarship in the Humanities},
	publisher = {Oxford University Press},
	author = {Stopponi, Silvia and Peels-Matthey, Saskia and Nissim, Malvina},
	year = {2024},
}

@article{dexter_database_2024,
	title = {A {Database} of {Intertexts} in {Valerius} {Flaccus}' {Argonautica}: {A} {Benchmarking} {Resource} for the {Evaluation} of {Computational} {Intertextual} {Search} of {Latin} {Corpora}},
	volume = {10},
	issn = {2059-481X},
	shorttitle = {\“{A} {Database} of {Intertexts} in {Valerius} {Flaccus}\’ {Argonautica}\ 1},
	url = {https://openhumanitiesdata.metajnl.com/articles/10.5334/johd.153},
	doi = {10.5334/johd.153},
	abstract = {Characterization of intertextual references among authors is fundamental for the study of Latin literature. In this paper, we describe a large-scale intertextuality dataset compiled from three modern commentaries on Valerius Flaccus\’ epic poem Argonautica. The dataset includes 945 references to earlier and contemporary Roman authors, as well as associated metadata required for use of multiple intertext search tools. To illustrate the dataset\’s reuse potential, we perform a new benchmark analysis of Fīlum, a sequence alignment tool for intertextuality detection.},
	language = {en-US},
	number = {1},
	urldate = {2025-07-31},
	journal = {Journal of Open Humanities Data},
	author = {Dexter, Joseph P. and Chaudhuri, Pramit and Burns, Patrick J. and Adams, Elizabeth D. and Bolt, Thomas J. and Cásarez, Adriana and Flynt, Jeffrey H. and Li, Kyle and Patterson, James F. and Schwartz, Ariane and Shumway, Scott},
	month = jan,
	year = {2024},
}

Characterization of intertextual references among authors is fundamental for the study of Latin literature. In this paper, we describe a large-scale intertextuality dataset compiled from three modern commentaries on Valerius Flaccus’ epic poem Argonautica. The dataset includes 945 references to earlier and contemporary Roman authors, as well as associated metadata required for use of multiple intertext search tools. To illustrate the dataset’s reuse potential, we perform a new benchmark analysis of Fīlum, a sequence alignment tool for intertextuality detection.

@inproceedings{dreyer_digital_2024,
	address = {Athens},
	title = {Digital {Sovereignty} – {Results} from the {ZKI} {Survey} in {DACH}},
	url = {https://easychair.org/publications/paper/bfg7},
	doi = {10.29007/zmwf},
	abstract = {The annual ZKI top trends survey for 2024 had a focus on Digital Sovereignty (DS) and asked about the views on aspects of DS, Open-Source-Software, Clouds, Collaboration Structures and Security. This article illustrates the survey results and discusses key issues and outcomes.},
	urldate = {2025-11-17},
	booktitle = {Proceedings of {EUNIS} 2024 annual congress in {Athens}},
	author = {Dreyer, Malte},
	year = {2024},
	pages = {180--171},
}

The annual ZKI top trends survey for 2024 had a focus on Digital Sovereignty (DS) and asked about the views on aspects of DS, Open-Source-Software, Clouds, Collaboration Structures and Security. This article illustrates the survey results and discusses key issues and outcomes.

@article{kuehnast_development_2024,
	title = {Development of basic reading skills in {Latin}: a corpus-based tool for computer-assisted fluency training},
	volume = {11},
	issn = {null},
	shorttitle = {Development of basic reading skills in {Latin}},
	url = {https://doi.org/10.1080/2331186X.2024.2416819},
	doi = {10.1080/2331186X.2024.2416819},
	abstract = {The present paper evaluates the processes of reading acquisition in Latin from the component-skills approach and discusses how advances in reading in modern foreign languages could be adapted to the specific needs of Latin as a historical language. Compared to the holistic and socially embedded approaches to modern foreign language acquisition, the grammar-translation method traditionally used in schools shows considerable weaknesses in the development of basic reading skills in Latin. Therefore, we address the possible advantages of corpus-based teaching strategies and present Machina Callida, a psycholinguistically informed e-tutor suitable for supporting Latin vocabulary acquisition and reading comprehension at beginner and intermediate levels. Using digital corpora of original Latin texts, the application semi-automatically generates contextualized vocabulary exercises tailored to the needs of different groups of learners. Through its integration with the research data repository Zenodo, Machina Callida supports online collaboration in the creation and distribution of open educational resources through crowdsourcing.},
	number = {1},
	urldate = {2024-10-22},
	journal = {Cogent Education},
	publisher = {Cogent OA},
	author = {Kuehnast, Milena and Schulz, Konstantin and Lüdeling, Anke},
	month = dec,
	year = {2024},
	note = {\_eprint: https://doi.org/10.1080/2331186X.2024.2416819},
	keywords = {CALL, Classical Language \& Literature, Classroom Practice, Databases, Language \& Linguistics, Language Teaching \& Learning, Latin, Open \& Distance Education and eLearning, Teaching \& Learning - Education, bold, corpus resources, reading comprehension, scientific, vocabulary acquisition},
	pages = {2416819},
}

The present paper evaluates the processes of reading acquisition in Latin from the component-skills approach and discusses how advances in reading in modern foreign languages could be adapted to the specific needs of Latin as a historical language. Compared to the holistic and socially embedded approaches to modern foreign language acquisition, the grammar-translation method traditionally used in schools shows considerable weaknesses in the development of basic reading skills in Latin. Therefore, we address the possible advantages of corpus-based teaching strategies and present Machina Callida, a psycholinguistically informed e-tutor suitable for supporting Latin vocabulary acquisition and reading comprehension at beginner and intermediate levels. Using digital corpora of original Latin texts, the application semi-automatically generates contextualized vocabulary exercises tailored to the needs of different groups of learners. Through its integration with the research data repository Zenodo, Machina Callida supports online collaboration in the creation and distribution of open educational resources through crowdsourcing.

@inproceedings{schulz_seflag_2024,
	address = {Miami, USA},
	title = {{SEFLAG}: {Systematic} {Evaluation} {Framework} for {NLP} {Models} and {Datasets} in {Latin} and {Ancient} {Greek}},
	shorttitle = {{SEFLAG}},
	url = {https://aclanthology.org/2024.nlp4dh-1.24},
	doi = {10.18653/v1/2024.nlp4dh-1.24},
	abstract = {Literary scholars of Latin and Ancient Greek increasingly use natural language processing for their work, but many models and datasets are hard to use due to a lack of sustainable research data management. This paper introduces the Systematic Evaluation Framework for natural language processing models and datasets in Latin and Ancient Greek (SEFLAG), which consistently assesses language resources using common criteria, such as specific evaluation metrics, metadata and risk analysis. The framework, a work in progress in its initial phase, currently covers lemmatization and named entity recognition for both languages, with plans for adding dependency parsing and other tasks. For increased transparency and sustainability, a thorough documentation is included as well as an integration into the HuggingFace ecosystem. The combination of these efforts is designed to support researchers in their search for suitable models.},
	urldate = {2024-11-12},
	booktitle = {Proceedings of the 4th {International} {Conference} on {Natural} {Language} {Processing} for {Digital} {Humanities}},
	publisher = {Association for Computational Linguistics},
	author = {Schulz, Konstantin and Deichsler, Florian},
	editor = {Hämäläinen, Mika and Öhman, Emily and Miyagawa, So and Alnajjar, Khalid and Bizzoni, Yuri},
	month = nov,
	year = {2024},
	keywords = {bold, scientific},
	pages = {247--258},
}

Literary scholars of Latin and Ancient Greek increasingly use natural language processing for their work, but many models and datasets are hard to use due to a lack of sustainable research data management. This paper introduces the Systematic Evaluation Framework for natural language processing models and datasets in Latin and Ancient Greek (SEFLAG), which consistently assesses language resources using common criteria, such as specific evaluation metrics, metadata and risk analysis. The framework, a work in progress in its initial phase, currently covers lemmatization and named entity recognition for both languages, with plans for adding dependency parsing and other tasks. For increased transparency and sustainability, a thorough documentation is included as well as an integration into the HuggingFace ecosystem. The combination of these efforts is designed to support researchers in their search for suitable models.

@inproceedings{luthra_data-envelopes_2024,
	address = {Torino, Italia},
	title = {Data-{Envelopes} for {Cultural} {Heritage}: {Going} beyond {Datasheets}.},
	url = {https://aclanthology.org/2024.legal-1.9/},
	booktitle = {Proceedings of the {Workshop} on {Legal} and {Ethical} {Issues} in {Human} {Language} {Technologies} @ {LREC}-{COLING} 2024},
	author = {Luthra, Mrinalini and Eskevich, Maria},
	year = {2024},
	pages = {52--65},
}

@book{beine_doppelte_2024,
	address = {Göttingen},
	series = {Studia {Comica}},
	title = {Das doppelte {Spiel} des "servus callidus": eine poetologische und gesellschaftliche {Reflexionsfigur} auf den europäischen {Bühnen} der {Frühen} {Neuzeit}},
	isbn = {978-3-911065-09-2 978-3-911065-11-5},
	shorttitle = {Das doppelte {Spiel} des "servus callidus"},
	doi = {10.13109/9783911065115},
	abstract = {In der römischen Komödie gestaltet der Figurentypus des intriganten Sklaven sein Intrigenspiel aus einer gesellschaftlich untergeordneten Position heraus mit einer dichterischen Schaffenskraft und setzt sich sowohl mit seinem poetischen als auch gesellschaftlichen Handlungsrahmen auseinander. Dem „servus callidus“ wohnt damit ein poetologisches und gesellschaftliches Reflexionspotential inne, das ihn losgelöst vom antiken Kontext über Jahrhunderte anschlussfähig macht. Die frühneuzeitlichen Dramatiker gestalten dieses doppelte Reflexionspotential bezogen auf zeitgenössische Diskurse neu aus. Ihre Auseinandersetzung mit den antiken Vorlagen schlägt sich in Form zahlreicher intertextueller Verweise in den Komödien nieder. Über diese intertextuellen Bezüge untersucht die Arbeit die Rezeption des „servus callidus“ anhand ausgewählter Beispiele aus dem englischen, französischen und neulateinischen Theater. Ergänzend zur inhaltlichen Textanalyse eröffnet die digitale Methode der Netzwerkanalyse eine dramenübergreifende Perspektive auf diesen Figurentypus. Aufgrund ihrer interdisziplinären Ausrichtung sowie einstiegsfreundlichen Konzeption richtet sich die Arbeit sowohl an ein breitgefächertes wissenschaftliches Fachpublikum als auch an alle Interessierten des europäischen Theaters},
	language = {ger},
	number = {Band 23},
	publisher = {Verlag Antike},
	author = {Beine, Julia Jennifer},
	year = {2024},
}

In der römischen Komödie gestaltet der Figurentypus des intriganten Sklaven sein Intrigenspiel aus einer gesellschaftlich untergeordneten Position heraus mit einer dichterischen Schaffenskraft und setzt sich sowohl mit seinem poetischen als auch gesellschaftlichen Handlungsrahmen auseinander. Dem „servus callidus“ wohnt damit ein poetologisches und gesellschaftliches Reflexionspotential inne, das ihn losgelöst vom antiken Kontext über Jahrhunderte anschlussfähig macht. Die frühneuzeitlichen Dramatiker gestalten dieses doppelte Reflexionspotential bezogen auf zeitgenössische Diskurse neu aus. Ihre Auseinandersetzung mit den antiken Vorlagen schlägt sich in Form zahlreicher intertextueller Verweise in den Komödien nieder. Über diese intertextuellen Bezüge untersucht die Arbeit die Rezeption des „servus callidus“ anhand ausgewählter Beispiele aus dem englischen, französischen und neulateinischen Theater. Ergänzend zur inhaltlichen Textanalyse eröffnet die digitale Methode der Netzwerkanalyse eine dramenübergreifende Perspektive auf diesen Figurentypus. Aufgrund ihrer interdisziplinären Ausrichtung sowie einstiegsfreundlichen Konzeption richtet sich die Arbeit sowohl an ein breitgefächertes wissenschaftliches Fachpublikum als auch an alle Interessierten des europäischen Theaters

@article{rachbauer_chancen_2024,
	title = {Chancen und {Potenziale} von {Künstlicher} {Intelligenz} ({KI}) und {Learning} {Analytics} ({LA}) auf inklusive {Begabungsförderung}},
	volume = {4},
	copyright = {https://creativecommons.org/licenses/by-nc-nd/4.0},
	issn = {2791-4046, 2791-4046},
	url = {https://schule-verantworten.education/journal/index.php/sv/article/view/390},
	doi = {10.53349/schuleverantworten.2024.i1.a390},
	abstract = {Der Beitrag betont den einschneidenden Einfluss von Künstlicher Intelligenz (KI) und Learning Analytics (LA) auf die Bildung. Ihr gezieltes Zusammenspiel verspricht nicht nur maßgeschneiderte Begabungsförderung, sondern auch eine grundlegende Transformation des Bildungswesens. Besonders im Kontext inklusiver Bildungsprozesse können KI und LA dazu beitragen, individualisierte Lernangebote zu schaffen und Heterogenität besser zu berücksichtigen. Lehrkräfte spielen eine zentrale Rolle und benötigen umfassende Weiterbildungen, sowohl technischer als auch ethischer Natur. Die effektive Integration dieser Technologien erfordert nicht nur technische Infrastruktur, sondern auch kompetente Pädagog*innen. Die Zukunft sollte sich auf die Weiterentwicklung von KI- und LA-Systemen sowie auf die Ausbildung der Lehrkräfte in diesem Bereich konzentrieren, unter Berücksichtigung sozialer, ethischer und rechtlicher Aspekte.},
	number = {1},
	urldate = {2024-09-01},
	journal = {\#schuleverantworten},
	author = {Rachbauer, Tamara},
	month = mar,
	year = {2024},
}

Der Beitrag betont den einschneidenden Einfluss von Künstlicher Intelligenz (KI) und Learning Analytics (LA) auf die Bildung. Ihr gezieltes Zusammenspiel verspricht nicht nur maßgeschneiderte Begabungsförderung, sondern auch eine grundlegende Transformation des Bildungswesens. Besonders im Kontext inklusiver Bildungsprozesse können KI und LA dazu beitragen, individualisierte Lernangebote zu schaffen und Heterogenität besser zu berücksichtigen. Lehrkräfte spielen eine zentrale Rolle und benötigen umfassende Weiterbildungen, sowohl technischer als auch ethischer Natur. Die effektive Integration dieser Technologien erfordert nicht nur technische Infrastruktur, sondern auch kompetente Pädagog*innen. Die Zukunft sollte sich auf die Weiterentwicklung von KI- und LA-Systemen sowie auf die Ausbildung der Lehrkräfte in diesem Bereich konzentrieren, unter Berücksichtigung sozialer, ethischer und rechtlicher Aspekte.

@misc{european_parliament_regulation_2024,
	title = {Regulation ({EU}) 2024/1689 laying down harmonised rules on artificial intelligence},
	shorttitle = {{EU} {AI} {Act}},
	url = {https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng},
	language = {en},
	urldate = {2025-08-18},
	journal = {EUR-Lex},
	author = {European Parliament and European Council},
	month = jun,
	year = {2024},
	note = {Doc ID: 32024R1689
Doc Sector: 3
Doc Title: Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA relevance)
Doc Type: R
Usr\_lan: en},
}

@misc{schulz_kunstliche_2024,
	address = {Berlin},
	title = {Künstliche {Intelligenz} in der {Sprachverarbeitung}},
	url = {https://zenodo.org/records/11190250},
	doi = {10.5281/zenodo.11190250},
	abstract = {Folien zu einem Vortrag},
	language = {deu},
	urldate = {2024-05-14},
	author = {Schulz, Konstantin},
	month = may,
	year = {2024},
	keywords = {Computer and information sciences, Languages and literature, Linguistics, other},
}

Folien zu einem Vortrag

@article{beyer_rezension_2024,
	series = {Latein und {Griechisch} in {Berlin} und {Brandenburg}},
	title = {Rezension zu „{KI}-{Bildung} im {Lateinunterricht} – {Ein} schulpraktischer {Leitfaden}“ der {Reihe} {KI}-{Bildung} im {Ovid} {Verlag}, {Hrsg}. {Rudolf} {Henneböhl}},
	volume = {68},
	issn = {0945-2257},
	url = {https://davbb.de/sammlung-ausgaben-lgbb/lgbb-2-2024/},
	doi = {https://davbb.de/sammlung-ausgaben-lgbb/lgbb-2-2024/},
	number = {2},
	journal = {LGBB},
	author = {Beyer, Andrea},
	year = {2024},
	keywords = {other},
	pages = {148--152},
}

@article{taha_comprehensive_2024,
	title = {A comprehensive survey of text classification techniques and their research applications: {Observational} and experimental insights},
	volume = {54},
	issn = {1574-0137},
	shorttitle = {A comprehensive survey of text classification techniques and their research applications},
	url = {https://www.sciencedirect.com/science/article/pii/S1574013724000480},
	doi = {10.1016/j.cosrev.2024.100664},
	abstract = {The exponential growth of textual data presents substantial challenges in management and analysis, notably due to high storage and processing costs. Text classification, a vital aspect of text mining, provides robust solutions by enabling efficient categorization and organization of text data. These techniques allow individuals, researchers, and businesses to derive meaningful patterns and insights from large volumes of text. This survey paper introduces a comprehensive taxonomy specifically designed for text classification based on research fields. The taxonomy is structured into hierarchical levels: research field-based category, research field-based sub-category, methodology-based technique, methodology sub-technique, and research field applications. We employ a dual evaluation approach: empirical and experimental. Empirically, we assess text classification techniques across four critical criteria. Experimentally, we compare and rank the methodology sub-techniques within the same methodology technique and within the same overall research field sub-category. This structured taxonomy, coupled with thorough evaluations, provides a detailed and nuanced understanding of text classification algorithms and their applications, empowering researchers to make informed decisions based on precise, field-specific insights.},
	urldate = {2025-07-07},
	journal = {Computer Science Review},
	author = {Taha, Kamal and Yoo, Paul D. and Yeun, Chan and Homouz, Dirar and Taha, Aya},
	month = nov,
	year = {2024},
	keywords = {Artificial intelligence, Data science, Deep learning, Text classification, Text data mining},
	pages = {100664},
}

The exponential growth of textual data presents substantial challenges in management and analysis, notably due to high storage and processing costs. Text classification, a vital aspect of text mining, provides robust solutions by enabling efficient categorization and organization of text data. These techniques allow individuals, researchers, and businesses to derive meaningful patterns and insights from large volumes of text. This survey paper introduces a comprehensive taxonomy specifically designed for text classification based on research fields. The taxonomy is structured into hierarchical levels: research field-based category, research field-based sub-category, methodology-based technique, methodology sub-technique, and research field applications. We employ a dual evaluation approach: empirical and experimental. Empirically, we assess text classification techniques across four critical criteria. Experimentally, we compare and rank the methodology sub-techniques within the same methodology technique and within the same overall research field sub-category. This structured taxonomy, coupled with thorough evaluations, provides a detailed and nuanced understanding of text classification algorithms and their applications, empowering researchers to make informed decisions based on precise, field-specific insights.

@article{calleja_ibanez_exploring_2024,
	title = {Exploring named-entity recognition techniques for academic books},
	volume = {37},
	copyright = {© 2024 The Author(s). Learned Publishing published by John Wiley \& Sons Ltd on behalf of ALPSP.},
	issn = {1741-4857},
	url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/leap.1610},
	doi = {10.1002/leap.1610},
	abstract = {Recent advances in the natural language processing (NLP) field have achieved impressive results in various tasks. However, NLP techniques are underrepresented in the analysis of Humanities and Social Science texts and in languages other than English. In particular, academic books are a highly valuable source of information that has not been exploited by these techniques at all. The recognition of named entities (person names, organizations or locations) and their semantic annotation over books could enrich the visibility and discoverability of the information by users. This is an opportunity for academia and the academic publishing industry in which semantic search is a central task and now books can be queried by named entities of interest that are in their content. This work proposes a methodology to apply named-entity recognition to publish the results into an ontological semantic-web format. The work has been performed over a corpus of academic books provided by UNE (Unión de Editoriales Universitarias Españolas, Union of Spanish University Presses). Results show an enrichment of the information extracted over the books and of the possibilities of querying them at the individual level but also within the whole set of books, increasing the possibilities for books to be discovered or retrieved beyond metadata.},
	language = {en},
	number = {3},
	urldate = {2025-07-07},
	journal = {Learned Publishing},
	author = {Calleja Ibañez, Pablo and Giménez-Toledo, Elea},
	year = {2024},
	note = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/leap.1610},
	keywords = {academic books, discoverability, multilingualism, name entity recognition (NER), onomastic index, ontology, semantic-web},
	pages = {e1610},
}

Recent advances in the natural language processing (NLP) field have achieved impressive results in various tasks. However, NLP techniques are underrepresented in the analysis of Humanities and Social Science texts and in languages other than English. In particular, academic books are a highly valuable source of information that has not been exploited by these techniques at all. The recognition of named entities (person names, organizations or locations) and their semantic annotation over books could enrich the visibility and discoverability of the information by users. This is an opportunity for academia and the academic publishing industry in which semantic search is a central task and now books can be queried by named entities of interest that are in their content. This work proposes a methodology to apply named-entity recognition to publish the results into an ontological semantic-web format. The work has been performed over a corpus of academic books provided by UNE (Unión de Editoriales Universitarias Españolas, Union of Spanish University Presses). Results show an enrichment of the information extracted over the books and of the possibilities of querying them at the individual level but also within the whole set of books, increasing the possibilities for books to be discovered or retrieved beyond metadata.

@misc{schulz_seflag_2024,
	title = {{SEFLAG}: {Systematic} {Evaluation} {Framework} for {NLP} {Models} and {Datasets} in {Latin} and {Ancient} {Greek}},
	shorttitle = {{SEFLAG}},
	url = {https://zenodo.org/records/15790925},
	doi = {10.5281/zenodo.15790925},
	abstract = {The poster presents SEFLAG, a Systematic Evaluation Framework for NLP models and datasets in Latin and Ancient Greek, developed at Humboldt-Universität zu Berlin. It addresses three core research questions: helping literary scholars select suitable NLP models, systematically documenting language resources, and unifying similar but distinct annotation schemas.

SEFLAG integrates components such as model evaluation, data curation, and documentation, utilizing tools like spaCy, flair, Hugging Face, and Zenodo. It supports tasks including named entity recognition, lemmatization, and dependency parsing. A key feature is mapping between different annotation schemas to ensure comparability across resources. Evaluation metrics (such as F1, accuracy) show performance results for both Latin and Ancient Greek across several models and datasets.

Challenges in this domain include linguistic variation, limited resources, interoperability issues, and the need for sustainable, interdisciplinary research. SEFLAG contributes solutions such as publishing model cards and datasheets, using Linked Data for evaluation results, and offering case-specific mappings.

Future plans include expanding to more tasks, models, and datasets, creating educational materials on NLP evaluation, and fully integrating the framework into the Daidalos research infrastructure. All resources are open-access, with code and evaluation data available online.},
	language = {eng},
	urldate = {2025-07-02},
	author = {Schulz, Konstantin and Deichsler, Florian},
	month = nov,
	year = {2024},
	keywords = {Artificial Intelligence, Artificial intelligence, Languages and literature, Literature, Literature studies, Literature study, Natural Language Processing, Natural language processing},
}

The poster presents SEFLAG, a Systematic Evaluation Framework for NLP models and datasets in Latin and Ancient Greek, developed at Humboldt-Universität zu Berlin. It addresses three core research questions: helping literary scholars select suitable NLP models, systematically documenting language resources, and unifying similar but distinct annotation schemas. SEFLAG integrates components such as model evaluation, data curation, and documentation, utilizing tools like spaCy, flair, Hugging Face, and Zenodo. It supports tasks including named entity recognition, lemmatization, and dependency parsing. A key feature is mapping between different annotation schemas to ensure comparability across resources. Evaluation metrics (such as F1, accuracy) show performance results for both Latin and Ancient Greek across several models and datasets. Challenges in this domain include linguistic variation, limited resources, interoperability issues, and the need for sustainable, interdisciplinary research. SEFLAG contributes solutions such as publishing model cards and datasheets, using Linked Data for evaluation results, and offering case-specific mappings. Future plans include expanding to more tasks, models, and datasets, creating educational materials on NLP evaluation, and fully integrating the framework into the Daidalos research infrastructure. All resources are open-access, with code and evaluation data available online.

@article{potts_women_2024,
	title = {Women in {Greek} {Tragedy}: {Progression} of the {Euripidean} {Female} through the {Medium} of {Social} {Capital}},
	volume = {12},
	url = {https://www.scirp.org/pdf/als_2024013013440690.pdf},
	number = {1},
	journal = {Advances in Literary Study},
	publisher = {Scientific Research Publishing},
	author = {Potts, Craig and Philip, Susan},
	year = {2024},
	pages = {46--62},
}

@inproceedings{martinelli_exploring_2024,
	title = {Exploring {Neural} {Topic} {Modeling} on a {Classical} {Latin} {Corpus}},
	url = {https://aclanthology.org/2024.lrec-main.606.pdf},
	booktitle = {Proceedings of the 2024 {Joint} {International} {Conference} on {Computational} {Linguistics}, {Language} {Resources} and {Evaluation}},
	author = {Martinelli, Ginevra and Impicciché, Paola and Fersini, Elisabetta and Mambrini, Francesco and Passarotti, Marco},
	year = {2024},
	pages = {6929--6934},
}

@misc{sankarapu_dlbacktrace_2024,
	title = {{DLBacktrace}: {A} {Model} {Agnostic} {Explainability} for any {Deep} {Learning} {Models}},
	shorttitle = {{DLBacktrace}},
	url = {http://arxiv.org/abs/2411.12643},
	doi = {10.48550/arXiv.2411.12643},
	abstract = {The rapid advancement of artificial intelligence has led to increasingly sophisticated deep learning models, which frequently operate as opaque 'black boxes' with limited transparency in their decision-making processes. This lack of interpretability presents considerable challenges, especially in high-stakes applications where understanding the rationale behind a model's outputs is as essential as the outputs themselves. This study addresses the pressing need for interpretability in AI systems, emphasizing its role in fostering trust, ensuring accountability, and promoting responsible deployment in mission-critical fields. To address the interpretability challenge in deep learning, we introduce DLBacktrace, an innovative technique developed by the AryaXAI team to illuminate model decisions across a wide array of domains, including simple Multi Layer Perceptron (MLPs), Convolutional Neural Networks (CNNs), Large Language Models (LLMs), Computer Vision Models, and more. We provide a comprehensive overview of the DLBacktrace algorithm and present benchmarking results, comparing its performance against established interpretability methods, such as SHAP, LIME, GradCAM, Integrated Gradients, SmoothGrad, and Attention Rollout, using diverse task-based metrics. The proposed DLBacktrace technique is compatible with various model architectures built in PyTorch and TensorFlow, supporting models like Llama 3.2, other NLP architectures such as BERT and LSTMs, computer vision models like ResNet and U-Net, as well as custom deep neural network (DNN) models for tabular data. This flexibility underscores DLBacktrace's adaptability and effectiveness in enhancing model transparency across a broad spectrum of applications. The library is open-sourced and available at https://github.com/AryaXAI/DLBacktrace .},
	urldate = {2025-01-26},
	publisher = {arXiv},
	author = {Sankarapu, Vinay Kumar and Chitroda, Chintan and Rathore, Yashwardhan and Singh, Neeraj Kumar and Seth, Pratinav},
	month = nov,
	year = {2024},
	note = {arXiv:2411.12643 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Machine Learning},
}

The rapid advancement of artificial intelligence has led to increasingly sophisticated deep learning models, which frequently operate as opaque 'black boxes' with limited transparency in their decision-making processes. This lack of interpretability presents considerable challenges, especially in high-stakes applications where understanding the rationale behind a model's outputs is as essential as the outputs themselves. This study addresses the pressing need for interpretability in AI systems, emphasizing its role in fostering trust, ensuring accountability, and promoting responsible deployment in mission-critical fields. To address the interpretability challenge in deep learning, we introduce DLBacktrace, an innovative technique developed by the AryaXAI team to illuminate model decisions across a wide array of domains, including simple Multi Layer Perceptron (MLPs), Convolutional Neural Networks (CNNs), Large Language Models (LLMs), Computer Vision Models, and more. We provide a comprehensive overview of the DLBacktrace algorithm and present benchmarking results, comparing its performance against established interpretability methods, such as SHAP, LIME, GradCAM, Integrated Gradients, SmoothGrad, and Attention Rollout, using diverse task-based metrics. The proposed DLBacktrace technique is compatible with various model architectures built in PyTorch and TensorFlow, supporting models like Llama 3.2, other NLP architectures such as BERT and LSTMs, computer vision models like ResNet and U-Net, as well as custom deep neural network (DNN) models for tabular data. This flexibility underscores DLBacktrace's adaptability and effectiveness in enhancing model transparency across a broad spectrum of applications. The library is open-sourced and available at https://github.com/AryaXAI/DLBacktrace .

@article{keersmaekers_adapting_2024,
	title = {Adapting transformer models to morphological tagging of two highly inflectional languages: a case study on {Ancient} {Greek} and {Latin}},
	shorttitle = {Adapting transformer models to morphological tagging of two highly inflectional languages},
	url = {https://aclanthology.org/2024.ml4al-1.17},
	doi = {10.18653/v1/2024.ml4al-1.17},
	abstract = {Natural language processing for Greek and Latin, inflectional languages with small corpora, requires special techniques. For morphological tagging, transformer models show promising potential, but the best approach to use these models is unclear. For both languages, this paper examines the impact of using morphological lexica, training different model types (a single model with a combined feature tag, multiple models for separate features, and a multi-task model for all features), and adding linguistic constraints. We find that, although simply fine-tuning transformers to predict a monolithic tag may already yield decent results, each of these adaptations can further improve tagging accuracy.},
	language = {en},
	urldate = {2025-01-26},
	journal = {Proceedings of the 1st Workshop on Machine Learning for Ancient Languages (ML4AL 2024)},
	publisher = {Association for Computational Linguistics},
	author = {Keersmaekers, Alek and Mercelis, Wouter},
	year = {2024},
	note = {Conference Name: Proceedings of the 1st Workshop on Machine Learning for Ancient Languages (ML4AL 2024)
Place: Hybrid in Bangkok, Thailand and online},
	pages = {165--176},
}

Natural language processing for Greek and Latin, inflectional languages with small corpora, requires special techniques. For morphological tagging, transformer models show promising potential, but the best approach to use these models is unclear. For both languages, this paper examines the impact of using morphological lexica, training different model types (a single model with a combined feature tag, multiple models for separate features, and a multi-task model for all features), and adding linguistic constraints. We find that, although simply fine-tuning transformers to predict a monolithic tag may already yield decent results, each of these adaptations can further improve tagging accuracy.

@article{soffiantini_cross-linguistic_2024,
	title = {Cross-linguistic annotation transfer in geoparsing experiments with {Classical} texts},
	volume = {6},
	url = {https://journal.dhbenelux.org/wp-content/uploads/2024/11/9_Soffiantini_individual.pdf},
	abstract = {The Natural History is an encyclopedic work written by the Latin author Pliny the Elder (first century CE). In this extensive text in 37 books, geography plays a pivotal role, with hundreds of mentions of ancient place names. In this paper, a geoparsing experiment is conducted on the Natural History with the scope of automatically identifying and extracting place entities. To achieve this, we take advantage of state-of-the-art NLP models to develop a multistage pipeline involving English Named Entity Recognition, English-Latin sentence alignment, and entity projection. The paper demonstrates how cross-lingual annotation transfer can be applied from a translation in a modern language back to the original text in the context of low-/medium-resource languages, such as Latin. The efficacy of the proposed pipeline is evaluated through the use of both standard metrics and a comprehensive manual error analysis. Additionally, the results are compared to those obtained by other Latin NER tools. Both analyses demonstrate that the proposed methodology achieves a superior f1-score. Finally, the majority of place entities were automatically associated with unique identifiers that enable geolocation by the projection of pre-disambiguated annotations derived from another geo-spatial project.},
	urldate = {2025-01-26},
	journal = {DH Benelux Journal},
	author = {Soffiantini, Laura},
	year = {2024},
	pages = {155--168},
}

The Natural History is an encyclopedic work written by the Latin author Pliny the Elder (first century CE). In this extensive text in 37 books, geography plays a pivotal role, with hundreds of mentions of ancient place names. In this paper, a geoparsing experiment is conducted on the Natural History with the scope of automatically identifying and extracting place entities. To achieve this, we take advantage of state-of-the-art NLP models to develop a multistage pipeline involving English Named Entity Recognition, English-Latin sentence alignment, and entity projection. The paper demonstrates how cross-lingual annotation transfer can be applied from a translation in a modern language back to the original text in the context of low-/medium-resource languages, such as Latin. The efficacy of the proposed pipeline is evaluated through the use of both standard metrics and a comprehensive manual error analysis. Additionally, the results are compared to those obtained by other Latin NER tools. Both analyses demonstrate that the proposed methodology achieves a superior f1-score. Finally, the majority of place entities were automatically associated with unique identifiers that enable geolocation by the projection of pre-disambiguated annotations derived from another geo-spatial project.

@article{beersmans_gotta_2024,
	title = {“{Gotta} catch ‘em all!”: {Retrieving} people in {Ancient} {Greek} texts combining transformer models and domain knowledge},
	shorttitle = {“{Gotta} catch ‘em all!”},
	url = {https://aclanthology.org/2024.ml4al-1.16.pdf},
	doi = {10.18653/v1/2024.ml4al-1.16},
	abstract = {In this paper, we present a study of transformer-based Named Entity Recognition (NER) as applied to Ancient Greek texts, with an emphasis on retrieving personal names. Recent research shows that, while the task remains difficult, the use of transformer models results in significant improvements. We, therefore, compare the performance of four transformer models on the task of NER for the categories of people, locations and groups, and add an out-of-domain test set to the existing datasets. Results on this set highlight the shortcomings of the models when confronted with a random sample of sentences. To be able to more straightforwardly integrate domain and linguistic knowledge to improve performance, we narrow down our approach to the category of people. The task is simplified to a binary PERS/MISC classification on the token level, starting from capitalised words. Next, we test the use of domain and linguistic knowledge to improve the results. We find that including simple gazetteer information as a binary mask has a marginally positive effect on newly annotated data and that treebanks can be used to help identify multi-word individuals if they are scarcely or inconsistently annotated in the available training data. The qualitative error analysis identifies the potential for improvement in both manual annotation and the inclusion of domain and linguistic knowledge in the transformer models.},
	language = {en},
	urldate = {2025-01-26},
	journal = {Proceedings of the 1st Workshop on Machine Learning for Ancient Languages (ML4AL 2024)},
	publisher = {Association for Computational Linguistics},
	author = {Beersmans, Marijke and Keersmaekers, Alek and De Graaf, Evelien and Van De Cruys, Tim and Depauw, Mark and Fantoli, Margherita},
	year = {2024},
	note = {Conference Name: Proceedings of the 1st Workshop on Machine Learning for Ancient Languages (ML4AL 2024)
Place: Hybrid in Bangkok, Thailand and online},
	pages = {152--164},
}

In this paper, we present a study of transformer-based Named Entity Recognition (NER) as applied to Ancient Greek texts, with an emphasis on retrieving personal names. Recent research shows that, while the task remains difficult, the use of transformer models results in significant improvements. We, therefore, compare the performance of four transformer models on the task of NER for the categories of people, locations and groups, and add an out-of-domain test set to the existing datasets. Results on this set highlight the shortcomings of the models when confronted with a random sample of sentences. To be able to more straightforwardly integrate domain and linguistic knowledge to improve performance, we narrow down our approach to the category of people. The task is simplified to a binary PERS/MISC classification on the token level, starting from capitalised words. Next, we test the use of domain and linguistic knowledge to improve the results. We find that including simple gazetteer information as a binary mask has a marginally positive effect on newly annotated data and that treebanks can be used to help identify multi-word individuals if they are scarcely or inconsistently annotated in the available training data. The qualitative error analysis identifies the potential for improvement in both manual annotation and the inclusion of domain and linguistic knowledge in the transformer models.

@inproceedings{palladino_development_2024,
	title = {Development of {Robust} {NER} {Models} and {Named} {Entity} {Tagsets} for {Ancient} {Greek}},
	url = {https://aclanthology.org/2024.lt4hala-1.11.pdf},
	abstract = {This contribution presents a novel approach to the development and evaluation of transformer-based models for Named Entity Recognition and Classification in Ancient Greek texts. We trained two models with annotated datasets by consolidating potentially ambiguous entity types under a harmonized set of classes. Then, we tested their performance with out-of-domain texts, reproducing a real-world use case. Both models performed very well under these conditions, with the multilingual model being slightly superior on the monolingual one. In the conclusion, we emphasize current limitations due to the scarcity of high-quality annotated corpora and to the lack of cohesive annotation strategies for ancient languages.},
	urldate = {2025-01-26},
	author = {Palladino, Chiara and Yousef, Tariq},
	year = {2024},
}

This contribution presents a novel approach to the development and evaluation of transformer-based models for Named Entity Recognition and Classification in Ancient Greek texts. We trained two models with annotated datasets by consolidating potentially ambiguous entity types under a harmonized set of classes. Then, we tested their performance with out-of-domain texts, reproducing a real-world use case. Both models performed very well under these conditions, with the multilingual model being slightly superior on the monolingual one. In the conclusion, we emphasize current limitations due to the scarcity of high-quality annotated corpora and to the lack of cohesive annotation strategies for ancient languages.

@inproceedings{sullutrone_large_2024,
	title = {Large {Language} {Models} integration in {Digital} {Humanities}},
	url = {https://www.semanticscholar.org/paper/Large-Language-Models-integration-in-Digital-Sullutrone/8d127bc2f01fab76ea9fdad8706d7f58fd4b8308},
	abstract = {The exponential growth of available data to Digital Humanities (DH) has created an impending need for tools capable of analyzing and extracting information from multi-lingual historical documents. This paper explores the research directions of my PhD project: providing DH scholars with effective, efficient, and explainable tools based on recent advancements in Large Language Models (LLMs). Two are the main directions of investigation: Self-Improving LLMs applied to Text-to-SQL and Topic Modeling, with a focus on interacting with and augmenting existing DBMS; Knowledge Graph (KG) creation and integration to mitigate hallucination, improve transparency and reasoning in question-answering systems. At the heart of my research lies the Digital Maktaba (DM) project which seeks to create a digital library for assisting in the preservation and analysis of multicultural non-latin heritage documents using, among others, cutting edge techniques for Natural Language Processing (NLP) and Data Science. The DM objectives and ideals align with the ultimate goal of the PhD project: the creation of instruments capable of aiding human-data interaction and information extraction while keeping the user at the center of an ever-evolving system. These tools have the potential to revolutionize the way DH scholars interact with historical documents, leading to new insights and discoveries for the field at large.},
	urldate = {2025-01-15},
	author = {Sullutrone, Giovanni},
	year = {2024},
}

The exponential growth of available data to Digital Humanities (DH) has created an impending need for tools capable of analyzing and extracting information from multi-lingual historical documents. This paper explores the research directions of my PhD project: providing DH scholars with effective, efficient, and explainable tools based on recent advancements in Large Language Models (LLMs). Two are the main directions of investigation: Self-Improving LLMs applied to Text-to-SQL and Topic Modeling, with a focus on interacting with and augmenting existing DBMS; Knowledge Graph (KG) creation and integration to mitigate hallucination, improve transparency and reasoning in question-answering systems. At the heart of my research lies the Digital Maktaba (DM) project which seeks to create a digital library for assisting in the preservation and analysis of multicultural non-latin heritage documents using, among others, cutting edge techniques for Natural Language Processing (NLP) and Data Science. The DM objectives and ideals align with the ultimate goal of the PhD project: the creation of instruments capable of aiding human-data interaction and information extraction while keeping the user at the center of an ever-evolving system. These tools have the potential to revolutionize the way DH scholars interact with historical documents, leading to new insights and discoveries for the field at large.

@misc{beyer_daidalos-projekt_2024,
	title = {Daidalos-{Projekt} - {Entwicklung} einer {Infrastruktur} zum {Einsatz} von {Natural} {Language} {Processing} für {Forschende} der {Klassischen} {Philologie}},
	url = {https://zenodo.org/records/12635794},
	doi = {10.5281/zenodo.12635794},
	abstract = {Project proposal approved by the German Research Council as part of their Funding Programme "e-Research Technologies" : https://www.dfg.de/en/research-funding/funding-opportunities/programmes/infrastructure/lis/funding-opportunities/e-research-technologies},
	language = {deu},
	urldate = {2024-07-03},
	publisher = {Zenodo},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = jul,
	year = {2024},
	keywords = {Artificial intelligence, Classics, Computer and information sciences, Linguistics, Natural language processing},
}

Project proposal approved by the German Research Council as part of their Funding Programme "e-Research Technologies" : https://www.dfg.de/en/research-funding/funding-opportunities/programmes/infrastructure/lis/funding-opportunities/e-research-technologies

@misc{sun_lalaeval_2024,
	title = {{LalaEval}: {A} {Holistic} {Human} {Evaluation} {Framework} for {Domain}-{Specific} {Large} {Language} {Models}},
	shorttitle = {{LalaEval}},
	url = {http://arxiv.org/abs/2408.13338},
	doi = {10.48550/arXiv.2408.13338},
	abstract = {This paper introduces LalaEval, a holistic framework designed for the human evaluation of domain-specific large language models (LLMs). LalaEval proposes a comprehensive suite of end-to-end protocols that cover five main components including domain specification, criteria establishment, benchmark dataset creation, construction of evaluation rubrics, and thorough analysis and interpretation of evaluation outcomes. This initiative aims to fill a crucial research gap by providing a systematic methodology for conducting standardized human evaluations within specific domains, a practice that, despite its widespread application, lacks substantial coverage in the literature and human evaluation are often criticized to be less reliable due to subjective factors, so standardized procedures adapted to the nuanced requirements of specific domains or even individual organizations are in great need. Furthermore, the paper demonstrates the framework's application within the logistics industry, presenting domain-specific evaluation benchmarks, datasets, and a comparative analysis of LLMs for the logistics domain use, highlighting the framework's capacity to elucidate performance differences and guide model selection and development for domain-specific LLMs. Through real-world deployment, the paper underscores the framework's effectiveness in advancing the field of domain-specific LLM evaluation, thereby contributing significantly to the ongoing discussion on LLMs' practical utility and performance in domain-specific applications.},
	urldate = {2024-09-03},
	publisher = {arXiv},
	author = {Sun, Chongyan and Lin, Ken and Wang, Shiwei and Wu, Hulong and Fu, Chengfei and Wang, Zhen},
	month = aug,
	year = {2024},
	note = {arXiv:2408.13338 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Human-Computer Interaction},
}

This paper introduces LalaEval, a holistic framework designed for the human evaluation of domain-specific large language models (LLMs). LalaEval proposes a comprehensive suite of end-to-end protocols that cover five main components including domain specification, criteria establishment, benchmark dataset creation, construction of evaluation rubrics, and thorough analysis and interpretation of evaluation outcomes. This initiative aims to fill a crucial research gap by providing a systematic methodology for conducting standardized human evaluations within specific domains, a practice that, despite its widespread application, lacks substantial coverage in the literature and human evaluation are often criticized to be less reliable due to subjective factors, so standardized procedures adapted to the nuanced requirements of specific domains or even individual organizations are in great need. Furthermore, the paper demonstrates the framework's application within the logistics industry, presenting domain-specific evaluation benchmarks, datasets, and a comparative analysis of LLMs for the logistics domain use, highlighting the framework's capacity to elucidate performance differences and guide model selection and development for domain-specific LLMs. Through real-world deployment, the paper underscores the framework's effectiveness in advancing the field of domain-specific LLM evaluation, thereby contributing significantly to the ongoing discussion on LLMs' practical utility and performance in domain-specific applications.

@misc{liang_whats_2024,
	title = {What's documented in {AI}? {Systematic} {Analysis} of {32K} {AI} {Model} {Cards}},
	shorttitle = {What's documented in {AI}?},
	url = {http://arxiv.org/abs/2402.05160},
	doi = {10.48550/arXiv.2402.05160},
	abstract = {The rapid proliferation of AI models has underscored the importance of thorough documentation, as it enables users to understand, trust, and effectively utilize these models in various applications. Although developers are encouraged to produce model cards, it's not clear how much information or what information these cards contain. In this study, we conduct a comprehensive analysis of 32,111 AI model documentations on Hugging Face, a leading platform for distributing and deploying AI models. Our investigation sheds light on the prevailing model card documentation practices. Most of the AI models with substantial downloads provide model cards, though the cards have uneven informativeness. We find that sections addressing environmental impact, limitations, and evaluation exhibit the lowest filled-out rates, while the training section is the most consistently filled-out. We analyze the content of each section to characterize practitioners' priorities. Interestingly, there are substantial discussions of data, sometimes with equal or even greater emphasis than the model itself. To evaluate the impact of model cards, we conducted an intervention study by adding detailed model cards to 42 popular models which had no or sparse model cards previously. We find that adding model cards is moderately correlated with an increase weekly download rates. Our study opens up a new perspective for analyzing community norms and practices for model documentation through large-scale data science and linguistics analysis.},
	urldate = {2024-09-03},
	publisher = {arXiv},
	author = {Liang, Weixin and Rajani, Nazneen and Yang, Xinyu and Ozoani, Ezinwanne and Wu, Eric and Chen, Yiqun and Smith, Daniel Scott and Zou, James},
	month = feb,
	year = {2024},
	note = {arXiv:2402.05160 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Machine Learning, Computer Science - Software Engineering},
}

The rapid proliferation of AI models has underscored the importance of thorough documentation, as it enables users to understand, trust, and effectively utilize these models in various applications. Although developers are encouraged to produce model cards, it's not clear how much information or what information these cards contain. In this study, we conduct a comprehensive analysis of 32,111 AI model documentations on Hugging Face, a leading platform for distributing and deploying AI models. Our investigation sheds light on the prevailing model card documentation practices. Most of the AI models with substantial downloads provide model cards, though the cards have uneven informativeness. We find that sections addressing environmental impact, limitations, and evaluation exhibit the lowest filled-out rates, while the training section is the most consistently filled-out. We analyze the content of each section to characterize practitioners' priorities. Interestingly, there are substantial discussions of data, sometimes with equal or even greater emphasis than the model itself. To evaluate the impact of model cards, we conducted an intervention study by adding detailed model cards to 42 popular models which had no or sparse model cards previously. We find that adding model cards is moderately correlated with an increase weekly download rates. Our study opens up a new perspective for analyzing community norms and practices for model documentation through large-scale data science and linguistics analysis.

@misc{liu_automatic_2024,
	title = {Automatic {Generation} of {Model} and {Data} {Cards}: {A} {Step} {Towards} {Responsible} {AI}},
	shorttitle = {Automatic {Generation} of {Model} and {Data} {Cards}},
	url = {https://arxiv.org/abs/2405.06258v2},
	abstract = {In an era of model and data proliferation in machine learning/AI especially marked by the rapid advancement of open-sourced technologies, there arises a critical need for standardized consistent documentation. Our work addresses the information incompleteness in current human-generated model and data cards. We propose an automated generation approach using Large Language Models (LLMs). Our key contributions include the establishment of CardBench, a comprehensive dataset aggregated from over 4.8k model cards and 1.4k data cards, coupled with the development of the CardGen pipeline comprising a two-step retrieval process. Our approach exhibits enhanced completeness, objectivity, and faithfulness in generated model and data cards, a significant step in responsible AI documentation practices ensuring better accountability and traceability.},
	language = {en},
	urldate = {2024-09-03},
	journal = {arXiv.org},
	author = {Liu, Jiarui and Li, Wenkai and Jin, Zhijing and Diab, Mona},
	month = may,
	year = {2024},
}

In an era of model and data proliferation in machine learning/AI especially marked by the rapid advancement of open-sourced technologies, there arises a critical need for standardized consistent documentation. Our work addresses the information incompleteness in current human-generated model and data cards. We propose an automated generation approach using Large Language Models (LLMs). Our key contributions include the establishment of CardBench, a comprehensive dataset aggregated from over 4.8k model cards and 1.4k data cards, coupled with the development of the CardGen pipeline comprising a two-step retrieval process. Our approach exhibits enhanced completeness, objectivity, and faithfulness in generated model and data cards, a significant step in responsible AI documentation practices ensuring better accountability and traceability.

@article{stopponi_natural_2024,
	title = {Natural {Language} {Processing} for {Ancient} {Greek}: {Design}, advantages and challenges of language models},
	issn = {0176-4225},
	journal = {Diachronica},
	publisher = {John Benjamins Publishing Company Amsterdam/Philadelphia},
	author = {Stopponi, Silvia and Pedrazzini, Nilo and Peels-Matthey, Saskia and McGillivray, Barbara and Nissim, Malvina},
	year = {2024},
}

@inproceedings{stussi_part--speech_2024,
	title = {Part-of-{Speech} {Tagging} of 16th-{Century} {Latin} with {GPT}},
	url = {https://aclanthology.org/2024.latechclfl-1.18.pdf},
	booktitle = {Proceedings of the 8th {Joint} {SIGHUM} {Workshop} on {Computational} {Linguistics} for {Cultural} {Heritage}, {Social} {Sciences}, {Humanities} and {Literature} ({LaTeCH}-{CLfL} 2024)},
	author = {Stüssi, Elina and Ströbel, Phillip},
	year = {2024},
	pages = {196--206},
}

@misc{noauthor_digital_2024,
	type = {online resource},
	title = {Digital {Tools} for {Learning} {Ancient} {Greek} and {Latin} and {Guiding} {Phrases} for {Using} {Generative} {AI} in {Ancient} {Language} {Study}},
	url = {https://figshare.com/articles/online_resource/Digital_Tools_for_Learning_Ancient_Greek_and_Latin_and_Guiding_Phrases_for_Using_Generative_AI_in_Ancient_Language_Study/25391782/3},
	doi = {10.6084/m9.figshare.25391782.v3},
	abstract = {This document is a short introductory guide to the digital tools available for supporting the study of Ancient Greek and Latin. The first part of this guide is a list of our preferred digital tools for supporting Ancient Greek and Latin learning. This list is not exhaustive, but it does include a variety of generative AI tools and their ideal uses for supporting ancient language learning. The second part of this guide is a series of pre-prepared prompts which can be copy-pasted into a conversational AI tool to guide the conversation towards your expected learning level. Before using generative AI to support your studies, make sure to take a look at some of our instructional videos about the ethics of using generative AI. \#STOPandTHINKbeforeyouGENERATE},
	language = {en},
	urldate = {2024-05-21},
	journal = {figshare},
	publisher = {figshare},
	month = mar,
	year = {2024},
}

This document is a short introductory guide to the digital tools available for supporting the study of Ancient Greek and Latin. The first part of this guide is a list of our preferred digital tools for supporting Ancient Greek and Latin learning. This list is not exhaustive, but it does include a variety of generative AI tools and their ideal uses for supporting ancient language learning. The second part of this guide is a series of pre-prepared prompts which can be copy-pasted into a conversational AI tool to guide the conversation towards your expected learning level. Before using generative AI to support your studies, make sure to take a look at some of our instructional videos about the ethics of using generative AI. #STOPandTHINKbeforeyouGENERATE

@inproceedings{sprugnoli_overview_2024,
	title = {Overview of the {EvaLatin} 2024 evaluation campaign},
	url = {https://aclanthology.org/2024.lt4hala-1.21.pdf},
	booktitle = {Proceedings of the {Third} {Workshop} on {Language} {Technologies} for {Historical} and {Ancient} {Languages} ({LT4HALA})@ {LREC}-{COLING}-2024},
	author = {Sprugnoli, Rachele and Iurescia, Federica and Passarotti, Marco},
	year = {2024},
	pages = {190--197},
}

@misc{krause_graphannis_2024,
	title = {{graphANNIS}},
	copyright = {Apache-2.0},
	url = {https://github.com/korpling/graphANNIS},
	doi = {10.5281/zenodo.2598164},
	abstract = {This is a new backend implementation of the ANNIS linguistic search and visualization system.},
	urldate = {2024-02-27},
	author = {Krause, Thomas},
	month = feb,
	year = {2024},
}

This is a new backend implementation of the ANNIS linguistic search and visualization system.

@inproceedings{jung_wahl_2023,
	title = {Die {Wahl} der {Mittel} - {Jupyter}-{Notebooks} als {Forschungsinfrastruktur}},
	url = {https://api.semanticscholar.org/CorpusID:257858099},
	booktitle = {Jahrestagung des {Verbands} {Digital} {Humanities} im deutschsprachigen {Raum}},
	author = {Jung, Kerstin and Hein, Pascal and Blessing, Andreas and Hess, Jan and Kushnarenko, Volodymyr},
	year = {2023},
}

@incollection{brown_linking_2023,
	edition = {DGO - Digital original},
	title = {Linking {Out}: {The} {Long} {Now} of {Digital} {Humanities} {Infrastructures}},
	url = {http://www.jstor.org/stable/jj.17681834.23},
	abstract = {A great promise of infrastructure is that it will link us all together. Christine Borgman (2007), for example, views the “added value of linking” (117) as one of the greatest benefits of digital research infrastructure (DRI). In this context, linking often means a system’s ability to link data and documents. But work in critical infrastructure studies extends the notion of linking further to designate DRI’s ability to link not only data and documents but also technologies, communities, and modes of knowledge organization and management (Ribes and Finholt 2009). Like Liu (2016), we believe DH research infrastructure is best understood through},
	urldate = {2026-01-25},
	booktitle = {Future {Horizons}: {Canadian} {Digital} {Humanities}},
	publisher = {University of Ottawa Press},
	author = {Brown, Susan and Martin, Kim and Ivanov, Asen},
	year = {2023},
	pages = {315--348},
}

A great promise of infrastructure is that it will link us all together. Christine Borgman (2007), for example, views the “added value of linking” (117) as one of the greatest benefits of digital research infrastructure (DRI). In this context, linking often means a system’s ability to link data and documents. But work in critical infrastructure studies extends the notion of linking further to designate DRI’s ability to link not only data and documents but also technologies, communities, and modes of knowledge organization and management (Ribes and Finholt 2009). Like Liu (2016), we believe DH research infrastructure is best understood through

@article{besa_forschungskompetenz_2023,
	title = {Forschungskompetenz von {Lehramtsstudierenden} und {Studierenden} anderer {Fachrichtungen}},
	volume = {13},
	issn = {2190-6904},
	url = {https://doi.org/10.1007/s35834-023-00382-w},
	doi = {10.1007/s35834-023-00382-w},
	abstract = {Für die Professionalität von Lehrkräften spielt Forschungskompetenz eine entscheidende Rolle, um evidenzbasiert handeln zu können. Es gibt Hinweise auf eine geringere Ausprägung der Forschungskompetenz seitens Studierender, welche ein Lehramt anstreben. Parallel hierzu wird über die Wirkung verschiedener Persönlichkeitsmerkmale auf die beruflichen Fähigkeiten zukünftiger Lehrer*innen diskutiert. Die vorliegende Studie untersucht, ob Unterschiede in der Forschungskompetenz sowie den Persönlichkeitsmerkmalen zwischen Studierenden verschiedener Lehrämter und Studierenden anderer Fachrichtungen bestehen. Der Fokus dieser Studie liegt auf der Überprüfung des Zusammenhangs zwischen den Persönlichkeitsmerkmalen und der Forschungskompetenz. Hierzu wurden 351 Studierende (74,1 \% weiblich) verschiedener Studiengänge (65,2 \% Lehramtsstudierende) zu ihrer Forschungskompetenz und den Persönlichkeitsmerkmalen befragt. Mithilfe von multivariaten Varianzanalysen (MANOVA) wurden signifikante Unterschiede in der selbsteingeschätzten Forschungskompetenz zwischen Lehramtsstudierenden und Studierenden anderer Fachrichtungen sowie zwischen den Lehramtsstudierenden unterschiedlicher Schulformen gefunden. Bezüglich der Persönlichkeitsmerkmale ließen sich keine signifikanten Unterschiede berichten. Abschließend konnten mithilfe einer multiplen Regressionsanalyse die Persönlichkeitsmerkmale Offenheit und Gewissenhaftigkeit als signifikante Prädiktoren mit moderater Varianzaufklärung für die Forschungskompetenz ermittelt werden.},
	language = {de},
	number = {1},
	urldate = {2025-12-30},
	journal = {Zeitschrift für Bildungsforschung},
	author = {Besa, Kris-Stephen and Lüking, Sarah and Biehl, Anna-Lisa and Wilde, Matthias},
	month = apr,
	year = {2023},
	keywords = {Forschungskompetenz, Lehramtsstudierende, Personality traits, Persönlichkeitsmerkmale, Research competence, Student teachers, Students of other disciplines, Studierende anderer Fachrichtungen},
	pages = {55--74},
}

Für die Professionalität von Lehrkräften spielt Forschungskompetenz eine entscheidende Rolle, um evidenzbasiert handeln zu können. Es gibt Hinweise auf eine geringere Ausprägung der Forschungskompetenz seitens Studierender, welche ein Lehramt anstreben. Parallel hierzu wird über die Wirkung verschiedener Persönlichkeitsmerkmale auf die beruflichen Fähigkeiten zukünftiger Lehrer*innen diskutiert. Die vorliegende Studie untersucht, ob Unterschiede in der Forschungskompetenz sowie den Persönlichkeitsmerkmalen zwischen Studierenden verschiedener Lehrämter und Studierenden anderer Fachrichtungen bestehen. Der Fokus dieser Studie liegt auf der Überprüfung des Zusammenhangs zwischen den Persönlichkeitsmerkmalen und der Forschungskompetenz. Hierzu wurden 351 Studierende (74,1 % weiblich) verschiedener Studiengänge (65,2 % Lehramtsstudierende) zu ihrer Forschungskompetenz und den Persönlichkeitsmerkmalen befragt. Mithilfe von multivariaten Varianzanalysen (MANOVA) wurden signifikante Unterschiede in der selbsteingeschätzten Forschungskompetenz zwischen Lehramtsstudierenden und Studierenden anderer Fachrichtungen sowie zwischen den Lehramtsstudierenden unterschiedlicher Schulformen gefunden. Bezüglich der Persönlichkeitsmerkmale ließen sich keine signifikanten Unterschiede berichten. Abschließend konnten mithilfe einer multiplen Regressionsanalyse die Persönlichkeitsmerkmale Offenheit und Gewissenhaftigkeit als signifikante Prädiktoren mit moderater Varianzaufklärung für die Forschungskompetenz ermittelt werden.

@inproceedings{zyska_care_2023,
	address = {Toronto, Canada},
	title = {{CARE}: {Collaborative} {AI}-{Assisted} {Reading} {Environment}},
	shorttitle = {{CARE}},
	url = {https://aclanthology.org/2023.acl-demo.28},
	doi = {10.18653/v1/2023.acl-demo.28},
	abstract = {Recent years have seen impressive progress in AI-assisted writing, yet the developments in AI-assisted reading are lacking. We propose inline commentary as a natural vehicle for AIbased reading assistance, and present CARE: the first open integrated platform for the study of inline commentary and reading. CARE facilitates data collection for inline commentaries in a commonplace collaborative reading environment, and provides a framework for enhancing reading with NLP-based assistance, such as text classification, generation or question answering. The extensible behavioral logging allows unique insights into the reading and commenting behavior, and flexible configuration makes the platform easy to deploy in new scenarios. To evaluate CARE in action, we apply the platform in a user study dedicated to scholarly peer review. CARE facilitates the data collection and study of inline commentary in NLP, extrinsic evaluation of NLP assistance, and application prototyping. We invite the community to explore and build upon the open source implementation of CARE1.},
	language = {en},
	urldate = {2025-12-30},
	booktitle = {Proceedings of the 61st {Annual} {Meeting} of the {Association} for {Computational} {Linguistics} ({Volume} 3: {System} {Demonstrations})},
	publisher = {Association for Computational Linguistics},
	author = {Zyska, Dennis and Dycke, Nils and Buchmann, Jan and Kuznetsov, Ilia and Gurevych, Iryna},
	year = {2023},
	pages = {291--303},
}

Recent years have seen impressive progress in AI-assisted writing, yet the developments in AI-assisted reading are lacking. We propose inline commentary as a natural vehicle for AIbased reading assistance, and present CARE: the first open integrated platform for the study of inline commentary and reading. CARE facilitates data collection for inline commentaries in a commonplace collaborative reading environment, and provides a framework for enhancing reading with NLP-based assistance, such as text classification, generation or question answering. The extensible behavioral logging allows unique insights into the reading and commenting behavior, and flexible configuration makes the platform easy to deploy in new scenarios. To evaluate CARE in action, we apply the platform in a user study dedicated to scholarly peer review. CARE facilitates the data collection and study of inline commentary in NLP, extrinsic evaluation of NLP assistance, and application prototyping. We invite the community to explore and build upon the open source implementation of CARE1.

@misc{pollin_new_2023,
	title = {New {Ways} of {Creating} {Research} {Data}:  {Conversion} of {Unstructured} {Text} to {TEI} {XML} using {GPT} on the {Correspondence} of {Hugo} {Schuchardt} with a {Web} {Prototype} for {Prompt} {Engineering}. {FORGE} 2023. {Tübingen}},
	shorttitle = {New {Ways} of {Creating} {Research} {Data}},
	url = {https://zenodo.org/records/8425163},
	doi = {10.5281/zenodo.8425163},
	abstract = {This paper explores the use of prompt engineering to streamline the creation of humanities research data by converting unstructured correspondence texts into the TEI XML format. The approach optimizes language models such as GPT to produce accurate structured data while preserving context. The paper discusses the iterative refinement of the conversion process, challenges and potential solutions, and presents a user-friendly web prototype. Overall, prompt engineering shows potential for improving the efficiency of research data creation in the humanities.},
	urldate = {2025-12-28},
	author = {Pollin, Christopher and Steiner, Christian and Zach, Constantin},
	month = oct,
	year = {2023},
	keywords = {Digital Edition, Generative AI, Research Data},
}

This paper explores the use of prompt engineering to streamline the creation of humanities research data by converting unstructured correspondence texts into the TEI XML format. The approach optimizes language models such as GPT to produce accurate structured data while preserving context. The paper discusses the iterative refinement of the conversion process, challenges and potential solutions, and presents a user-friendly web prototype. Overall, prompt engineering shows potential for improving the efficiency of research data creation in the humanities.

@misc{beyer_digitalgestutzte_2023,
	address = {Freiburg},
	title = {Digitalgestützte {Textanalyse} in {Forschung} und {Lehre}},
	url = {https://zenodo.org/records/8388745},
	doi = {10.5281/zenodo.8388745},
	abstract = {Die digitale Textanalyse eröffnet neue Perspektiven für Forschung und Lehre im Bereich der Klassischen Philologie. Dieser Beitrag beleuchtet die Vielfalt digitaler Hilfsmittel und Methoden, die sowohl fachunabhängige als auch fachspezifische Anwendungen ermöglichen – von der Digitalisierung und Darstellung von Texten über die linguistische und literaturwissenschaftliche Analyse bis hin zur Visualisierung von Interpretationsergebnissen. Anhand konkreter Beispiele aus der antiken Rhetorik, wie der Gegenüberstellung der Reden Caesars und Catos in Sallusts Coniuratio Catilinae, wird gezeigt, wie digitale Tools, z.B. Machina Callida und ChatGPT, zur quantitativen und qualitativen Analyse sprachlicher Unterschiede beitragen. Die Ergebnisse verdeutlichen signifikante Unterschiede in Satzlänge, Satzstruktur und Stilmerkmalen, die auf unterschiedliche rhetorische Strategien hinweisen. Der Beitrag unterstreicht zudem die Bedeutung einer fachspezifischen Digital Literacy, die nicht nur technische Kompetenzen, sondern auch kritisches Denken, Methodenentwicklung und die Fähigkeit zur eigenständigen Forschung fördert.},
	language = {deu},
	urldate = {2024-10-31},
	author = {Beyer, Andrea},
	month = sep,
	year = {2023},
	keywords = {Digital Classics, Digital Humanities, Latin language, language learning, other},
}

Die digitale Textanalyse eröffnet neue Perspektiven für Forschung und Lehre im Bereich der Klassischen Philologie. Dieser Beitrag beleuchtet die Vielfalt digitaler Hilfsmittel und Methoden, die sowohl fachunabhängige als auch fachspezifische Anwendungen ermöglichen – von der Digitalisierung und Darstellung von Texten über die linguistische und literaturwissenschaftliche Analyse bis hin zur Visualisierung von Interpretationsergebnissen. Anhand konkreter Beispiele aus der antiken Rhetorik, wie der Gegenüberstellung der Reden Caesars und Catos in Sallusts Coniuratio Catilinae, wird gezeigt, wie digitale Tools, z.B. Machina Callida und ChatGPT, zur quantitativen und qualitativen Analyse sprachlicher Unterschiede beitragen. Die Ergebnisse verdeutlichen signifikante Unterschiede in Satzlänge, Satzstruktur und Stilmerkmalen, die auf unterschiedliche rhetorische Strategien hinweisen. Der Beitrag unterstreicht zudem die Bedeutung einer fachspezifischen Digital Literacy, die nicht nur technische Kompetenzen, sondern auch kritisches Denken, Methodenentwicklung und die Fähigkeit zur eigenständigen Forschung fördert.

@misc{beyer_chatbot_2023,
	address = {Berlin, Germany},
	title = {@chatbot: warum kannst du latein et quo vadis?},
	shorttitle = {@chatbot},
	url = {https://zenodo.org/record/8412875},
	doi = {10.5281/zenodo.8412875},
	abstract = {Der Vortrag untersucht die Fähigkeit von Large Language Models (LLMs) im Kontext der lateinischen Sprache. Er beleuchtet, wie LLMs durch das Training auf umfangreichen, multilingualen Korpora – darunter auch lateinische Texte – in der Lage sind, sowohl lateinische als auch deutsche Texte zu generieren und zu übersetzen. Die Analyse zeigt, dass LLMs aufgrund ihrer Fähigkeit, Muster in Wortverteilungen zu erkennen und Kohärenz über mehrere Absätze hinweg zu gewährleisten, in der Lage sind, authentisch erscheinende lateinische Texte zu erstellen, die auf Lehrbuchstil und grammatische Strukturen abgestimmt sind. Gleichzeitig werden Herausforderungen wie mögliche Verzerrungen in den Trainingsdaten, ethische Bedenken bei der Generierung von Inhalten sowie rechtliche Fragen im Hinblick auf Urheberrechte und die Nutzung geschützter Materialien diskutiert. An einem angeblich historischen Fund – einer von KI erzeugten lateinischen Dankesrede Caesars an Cicero – wird gezeigt, welche Gefahren durch die Fähigkeiten von KI entstehen und wie wichtig es daher ist, Fachwissen und kritisches Denken miteinander zu vereinen.},
	language = {deu},
	urldate = {2023-10-06},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = oct,
	year = {2023},
	keywords = {Latin language, artificial intelligence, artificial intelligence literacy, language learning, other},
}

Der Vortrag untersucht die Fähigkeit von Large Language Models (LLMs) im Kontext der lateinischen Sprache. Er beleuchtet, wie LLMs durch das Training auf umfangreichen, multilingualen Korpora – darunter auch lateinische Texte – in der Lage sind, sowohl lateinische als auch deutsche Texte zu generieren und zu übersetzen. Die Analyse zeigt, dass LLMs aufgrund ihrer Fähigkeit, Muster in Wortverteilungen zu erkennen und Kohärenz über mehrere Absätze hinweg zu gewährleisten, in der Lage sind, authentisch erscheinende lateinische Texte zu erstellen, die auf Lehrbuchstil und grammatische Strukturen abgestimmt sind. Gleichzeitig werden Herausforderungen wie mögliche Verzerrungen in den Trainingsdaten, ethische Bedenken bei der Generierung von Inhalten sowie rechtliche Fragen im Hinblick auf Urheberrechte und die Nutzung geschützter Materialien diskutiert. An einem angeblich historischen Fund – einer von KI erzeugten lateinischen Dankesrede Caesars an Cicero – wird gezeigt, welche Gefahren durch die Fähigkeiten von KI entstehen und wie wichtig es daher ist, Fachwissen und kritisches Denken miteinander zu vereinen.

@inproceedings{beyer_daidalos_2023,
	address = {Berlin},
	title = {{DAIdalos}: {Forschen} und {Lernen} zugleich?},
	isbn = {978-3-88579-731-9},
	shorttitle = {{DAIdalos}},
	url = {https://dl.gi.de/handle/20.500.12116/43162},
	doi = {10.18420/inf2023_42},
	abstract = {Die Daidalos-Infrastruktur entwickelt eine interaktive Lern- und Forschungsplattform für die Klassische Philologie und verwandte Disziplinen, die es Forschenden ermöglicht, Methoden des Natural Language Processing (NLP) an selbst zusammengestellten Forschungskorpora anzuwenden. Ausgerichtet am Ansatz des forschenden Lernens, integriert Daidalos authentische Forschungsfragen und den gesamten Workflow der Klassischen Philologie, um die Entwicklung von Data Literacy zu fördern. Die Plattform bietet mehrere Niveaustufen: von gesteuerten Demo-Workflows über domänenspezifische Interaktionen bis hin zur freien Konfiguration und Einbindung von eigenem Code. Dabei werden Kompetenzen wie Korpuszusammenstellung, Annotation, graphische Auswertung und Reflexion über den Einsatz von KI unterstützt. Daidalos verbindet so Forschung und Bildung, um Forschende niedrigschwellig an digitale Methoden heranzuführen und die fachspezifische KI-Bildung in der Klassischen Philologie voranzutreiben.},
	language = {de},
	urldate = {2023-12-14},
	publisher = {Gesellschaft für Informatik e.V.},
	author = {Beyer, Andrea and Schulz, Konstantin},
	year = {2023},
	keywords = {scientific},
	pages = {391--393},
}

Die Daidalos-Infrastruktur entwickelt eine interaktive Lern- und Forschungsplattform für die Klassische Philologie und verwandte Disziplinen, die es Forschenden ermöglicht, Methoden des Natural Language Processing (NLP) an selbst zusammengestellten Forschungskorpora anzuwenden. Ausgerichtet am Ansatz des forschenden Lernens, integriert Daidalos authentische Forschungsfragen und den gesamten Workflow der Klassischen Philologie, um die Entwicklung von Data Literacy zu fördern. Die Plattform bietet mehrere Niveaustufen: von gesteuerten Demo-Workflows über domänenspezifische Interaktionen bis hin zur freien Konfiguration und Einbindung von eigenem Code. Dabei werden Kompetenzen wie Korpuszusammenstellung, Annotation, graphische Auswertung und Reflexion über den Einsatz von KI unterstützt. Daidalos verbindet so Forschung und Bildung, um Forschende niedrigschwellig an digitale Methoden heranzuführen und die fachspezifische KI-Bildung in der Klassischen Philologie voranzutreiben.

@inproceedings{beyer_data_2023,
	address = {Tübingen, Germany},
	title = {Data {Literacy} für die {Klassische} {Philologie}: {dAIdalos} – eine interaktive {Infrastruktur} als {Lernangebot}},
	shorttitle = {Data {Literacy} für die {Klassische} {Philologie}},
	url = {https://zenodo.org/record/8420565},
	doi = {10.5281/zenodo.8420565},
	abstract = {Das DFG-geförderte explorative Projekt Daidalos entwickelt eine interaktive Forschungsinfrastruktur für die Klassische Philologie, die Forschenden den Zugriff auf Methoden des Natural Language Processing (NLP) an selbst zusammengestellten altsprachlichen Textkorpora ermöglicht. Ziel ist der Ausbau von Data Literacy durch forschungsorientierte, didaktisch gestaltete Lernbausteine, die die Zusammenstellung und Analyse von Korpora, den Umgang mit Annotierungen, TEI-XML und graphischen Auswertungen unterstützen. Mit Ansätzen der KI-Didaktik, forschenden Lernens und Micro-Learning werden authentische Beispiele, kurze Tutorien, kompetenzorientierte Übungen mit automatischem Feedback und individuelle Use-Case-Partnerschaften angeboten. Langfristig ist auf Basis eines entsprechend entwickelten Geschäftsmodells die Verankerung in einem universitären Rechenzentrum vorgesehen.},
	language = {deu},
	urldate = {2023-10-09},
	booktitle = {{FORGE} 2023 - {Forschungsdaten} in den {Geisteswissenschaften}: {Anything} {Goes}?! {Forschungsdaten} in den {Geisteswissenschaften} - kritisch betrachtet. {Konferenzabstracts}},
	publisher = {Zenodo},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2023},
	keywords = {Digital Classics, data literacy, open educational resources, research infrastructure, scientific},
}

Das DFG-geförderte explorative Projekt Daidalos entwickelt eine interaktive Forschungsinfrastruktur für die Klassische Philologie, die Forschenden den Zugriff auf Methoden des Natural Language Processing (NLP) an selbst zusammengestellten altsprachlichen Textkorpora ermöglicht. Ziel ist der Ausbau von Data Literacy durch forschungsorientierte, didaktisch gestaltete Lernbausteine, die die Zusammenstellung und Analyse von Korpora, den Umgang mit Annotierungen, TEI-XML und graphischen Auswertungen unterstützen. Mit Ansätzen der KI-Didaktik, forschenden Lernens und Micro-Learning werden authentische Beispiele, kurze Tutorien, kompetenzorientierte Übungen mit automatischem Feedback und individuelle Use-Case-Partnerschaften angeboten. Langfristig ist auf Basis eines entsprechend entwickelten Geschäftsmodells die Verankerung in einem universitären Rechenzentrum vorgesehen.

@misc{beyer_ki-bildung_2023,
	address = {Bielefeld, Germany},
	title = {{KI}-{Bildung}: {Was}, warum und wie?},
	shorttitle = {{KI}-{Bildung}},
	url = {https://zenodo.org/record/8381513},
	doi = {10.5281/zenodo.8381513},
	abstract = {Künstliche Intelligenz (KI) stellt eine transformative Kraft in der Bildung dar, die insbesondere im Fach Latein neue Herausforderungen und Chancen eröffnet. Dieser Vortrag beleuchtet die Bedeutung von KI-Bildung als zentralen Bestandteil digitaler Kompetenzen und erläutert ihre Kompetenzfelder: KI erkennen, verstehen, bewerten, anwenden und gestalten. Anhand des Beispiels der Rede des Critognatus aus Caesars Bellum Gallicum wird gezeigt, wie KI-Tools (Sprachmodelle) typische Aufgaben des Lateinunterrichts – etwa Übersetzen, Interpretieren und Präsentieren – effizient unterstützen können, jedoch mit systematischen Fehlern und Halluzinationen behaftet sind. Die Analyse unterstreicht die Notwendigkeit einer kritischen Auseinandersetzung mit KI-Ergebnissen, um die Qualität der Lernleistungen zu sichern. Im Kontext des Bildungsplans Latein wird aufgezeigt, wie KI-sensible Aufgabenformate – wie Reflexion über Übersetzungsentscheidungen, Vergleiche zwischen menschlichen und maschinellen Übersetzungen oder die kreative Weiterarbeit mit KI-Outputs – die fachliche und methodische Kompetenz der Lernenden stärken. Die Diskussion schließt mit konkreten Strategien für die Implementierung von KI-Bildung im Unterricht und verweist auf Ressourcen zur fachlichen Fortbildung von Lehrkräften.},
	language = {deu},
	urldate = {2023-09-29},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2023},
	keywords = {Latin language, artificial intelligence, artificial intelligence literacy, language learning, other},
}

Künstliche Intelligenz (KI) stellt eine transformative Kraft in der Bildung dar, die insbesondere im Fach Latein neue Herausforderungen und Chancen eröffnet. Dieser Vortrag beleuchtet die Bedeutung von KI-Bildung als zentralen Bestandteil digitaler Kompetenzen und erläutert ihre Kompetenzfelder: KI erkennen, verstehen, bewerten, anwenden und gestalten. Anhand des Beispiels der Rede des Critognatus aus Caesars Bellum Gallicum wird gezeigt, wie KI-Tools (Sprachmodelle) typische Aufgaben des Lateinunterrichts – etwa Übersetzen, Interpretieren und Präsentieren – effizient unterstützen können, jedoch mit systematischen Fehlern und Halluzinationen behaftet sind. Die Analyse unterstreicht die Notwendigkeit einer kritischen Auseinandersetzung mit KI-Ergebnissen, um die Qualität der Lernleistungen zu sichern. Im Kontext des Bildungsplans Latein wird aufgezeigt, wie KI-sensible Aufgabenformate – wie Reflexion über Übersetzungsentscheidungen, Vergleiche zwischen menschlichen und maschinellen Übersetzungen oder die kreative Weiterarbeit mit KI-Outputs – die fachliche und methodische Kompetenz der Lernenden stärken. Die Diskussion schließt mit konkreten Strategien für die Implementierung von KI-Bildung im Unterricht und verweist auf Ressourcen zur fachlichen Fortbildung von Lehrkräften.

@misc{beyer_mit_2023,
	address = {Bielefeld, Germany},
	title = {Mit und über {KI}-{Tools} im {Literaturunterricht} reflektieren},
	url = {https://zenodo.org/record/8388817},
	doi = {10.5281/zenodo.8388817},
	abstract = {Der Workshop präsentiert einen KI-sensiblen Lateinunterricht. Ziel ist es, Lernende zu befähigen, literarische Texte kritisch und reflektiert zu bearbeiten, wobei der Einsatz von KI-Tools wie ChatGPT explizit erlaubt und als unterstützende Arbeitsgrundlage vorgesehen ist. Die Lernaufgabe umfasst die strukturelle und inhaltliche Analyse der Rede, die Untersuchung rhetorischer Mittel und emotionaler Appelle, die situationsbezogene Übersetzung sowie die Diskussion der inneren Logik der Rede und Caesars Intentionen. Die Diskussion zeigt, dass KI-Tools nicht nur zur Unterstützung der Textanalyse dienen, sondern auch als Ausgangspunkt für kritische Reflexion über Stil, Intention und historische Perspektiven dienen können.},
	language = {deu},
	urldate = {2023-09-29},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2023},
	keywords = {Latin language, artificial intelligence, artificial intelligence literacy, language learning, other},
}

Der Workshop präsentiert einen KI-sensiblen Lateinunterricht. Ziel ist es, Lernende zu befähigen, literarische Texte kritisch und reflektiert zu bearbeiten, wobei der Einsatz von KI-Tools wie ChatGPT explizit erlaubt und als unterstützende Arbeitsgrundlage vorgesehen ist. Die Lernaufgabe umfasst die strukturelle und inhaltliche Analyse der Rede, die Untersuchung rhetorischer Mittel und emotionaler Appelle, die situationsbezogene Übersetzung sowie die Diskussion der inneren Logik der Rede und Caesars Intentionen. Die Diskussion zeigt, dass KI-Tools nicht nur zur Unterstützung der Textanalyse dienen, sondern auch als Ausgangspunkt für kritische Reflexion über Stil, Intention und historische Perspektiven dienen können.

@article{nguyen_tools_2023,
	title = {Tools for assessing teacher digital literacy: a review},
	issn = {2197-9995},
	shorttitle = {Tools for assessing teacher digital literacy},
	url = {https://doi.org/10.1007/s40692-022-00257-5},
	doi = {10.1007/s40692-022-00257-5},
	abstract = {With the rapid advancement of technology, digital literacy has become a key component in educators’ professional development. A wide range of assessment tools has been developed to measure teacher digital literacy; however, there has been no previous attempt to systematically synthesize and scrutinize those tools to improve evaluation of this ability among educators. The current study reviews literature on instruments that assess teacher digital literacy with the purposes of ascertaining the main aspects of it that recent researchers focus on in their evaluation, instrument types used for assessment, and the reliability and validity report, as well as the frameworks or models used to design assessment tools. The review selected 33 English-language publications in the field of educational technology from peer-reviewed journals indexed in the Education Resources Information Center (ERIC), Web of Science, and Scopus. The study period spanned from 2011 to 2022 with the objective of reviewing the tools used to assess teacher digital competence. The major findings demonstrate that scholars focus on digital competence in teachers’ use of educational technology, teaching and learning, professional development, and support for learners through digital competence. Other researchers emphasize the ability of educators to apply technology to the assessment of learner outcomes or to empower students in using technology to enhance learning. Additionally, self-evaluation instruments are common, whereas a few studies promote subjective evaluation in combination with objective assessment to provide a comprehensive understanding of teacher digital competence. The results form the basis for several recommendations for future research for the further examination of teacher digital literacy.},
	language = {en},
	urldate = {2023-09-14},
	journal = {Journal of Computers in Education},
	author = {Nguyen, Lan Anh Thuy and Habók, Anita},
	month = jan,
	year = {2023},
	keywords = {Assessment tools, Digital competence, Digital literacy, Educational assessment, Literature review, Professional development, Teacher evaluation},
}

With the rapid advancement of technology, digital literacy has become a key component in educators’ professional development. A wide range of assessment tools has been developed to measure teacher digital literacy; however, there has been no previous attempt to systematically synthesize and scrutinize those tools to improve evaluation of this ability among educators. The current study reviews literature on instruments that assess teacher digital literacy with the purposes of ascertaining the main aspects of it that recent researchers focus on in their evaluation, instrument types used for assessment, and the reliability and validity report, as well as the frameworks or models used to design assessment tools. The review selected 33 English-language publications in the field of educational technology from peer-reviewed journals indexed in the Education Resources Information Center (ERIC), Web of Science, and Scopus. The study period spanned from 2011 to 2022 with the objective of reviewing the tools used to assess teacher digital competence. The major findings demonstrate that scholars focus on digital competence in teachers’ use of educational technology, teaching and learning, professional development, and support for learners through digital competence. Other researchers emphasize the ability of educators to apply technology to the assessment of learner outcomes or to empower students in using technology to enhance learning. Additionally, self-evaluation instruments are common, whereas a few studies promote subjective evaluation in combination with objective assessment to provide a comprehensive understanding of teacher digital competence. The results form the basis for several recommendations for future research for the further examination of teacher digital literacy.

@article{sommerschield_machine_2023,
	title = {Machine learning for ancient languages: {A} survey},
	volume = {49},
	issn = {0891-2017},
	url = {https://direct.mit.edu/coli/article-pdf/doi/10.1162/coli_a_00481/2108936/coli_a_00481.pdf},
	doi = {10.1162/coli_a_00481},
	number = {3},
	journal = {Computational Linguistics},
	publisher = {MIT Press One Broadway, 12th Floor, Cambridge, Massachusetts 02142, USA …},
	author = {Sommerschield, Thea and Assael, Yannis and Pavlopoulos, John and Stefanak, Vanessa and Senior, Andrew and Dyer, Chris and Bodel, John and Prag, Jonathan and Androutsopoulos, Ion and Freitas, Nando de},
	year = {2023},
	pages = {703--747},
}

@misc{burns_latincy_2023,
	title = {{LatinCy}: {Synthetic} {Trained} {Pipelines} for {Latin} {NLP}},
	url = {https://arxiv.org/pdf/2305.04365.pdf},
	publisher = {arXiv},
	author = {Burns, Patrick J},
	year = {2023},
}

@article{palladino_translation_2023,
	title = {Translation {Alignment} for {Ancient} {Greek}: {Annotation} {Guidelines} and {Gold} {Standards}},
	volume = {9},
	issn = {2059-481X},
	shorttitle = {Translation {Alignment} for {Ancient} {Greek}},
	url = {https://openhumanitiesdata.metajnl.com/articles/10.5334/johd.131},
	doi = {10.5334/johd.131},
	abstract = {This paper covers three datasets containing texts in Ancient Greek, manually aligned at word level against translations in English (Grc-Eng), Portuguese (Grc-Por) and Latin (Grc-Lat). The datasets were collected by two domain experts through annotation on the Ugarit Translation Alignment Editor (https://ugarit.ialigner.com/). The quality of each dataset was measured through Inter-Annotator-Agreement (IAA) above 80\%. Each dataset contains the aligned pairs and an Annotation Style Guide, and serves as a Gold Standard for translation alignment of Ancient Greek, for the evaluation of automatic translation alignment models, and as high-quality training data. The Annotation Style Guide provides a starting point to approach the task of translation alignment for research and teaching. The data is stored on GitHub and Zenodo.},
	language = {en-US},
	number = {1},
	urldate = {2025-06-10},
	journal = {Journal of Open Humanities Data},
	author = {Palladino, Chiara and Shamsian, Farnoosh and Yousef, Tariq and Wright, David J. and Ferreira, Anise d’Orange and Reis, Michel Ferreira dos},
	month = nov,
	year = {2023},
}

This paper covers three datasets containing texts in Ancient Greek, manually aligned at word level against translations in English (Grc-Eng), Portuguese (Grc-Por) and Latin (Grc-Lat). The datasets were collected by two domain experts through annotation on the Ugarit Translation Alignment Editor (https://ugarit.ialigner.com/). The quality of each dataset was measured through Inter-Annotator-Agreement (IAA) above 80%. Each dataset contains the aligned pairs and an Annotation Style Guide, and serves as a Gold Standard for translation alignment of Ancient Greek, for the evaluation of automatic translation alignment models, and as high-quality training data. The Annotation Style Guide provides a starting point to approach the task of translation alignment for research and teaching. The data is stored on GitHub and Zenodo.

@article{alkemade_datasheets_2023,
	title = {Datasheets for {Digital} {Cultural} {Heritage} {Datasets}},
	volume = {19},
	number = {1},
	journal = {Journal of Open Humanities Data},
	author = {Alkemade, H. and Claeyssens, S. and Colavizza, G. and Freire, N. and Lehmann, J. and Neudecker, C. and Osti, G. and van Strien, D.},
	year = {2023},
	pages = {17},
}

@inproceedings{lorenz_ai-pack_2023,
	address = {Bonn},
	title = {{AI}-{PACK} – {Ein} {Rahmen} für {KI}-bezogene {Digitalkompetenzen} von {Lehrkräften} auf {Basis} von {DPACK}},
	url = {https://www.researchgate.net/publication/374350289_AI-PACK_-Ein_Rahmen_fur_KI-bezogene_Digitalkompetenzen_von_Lehrkraften_auf_Basis_von_DPACK},
	abstract = {Die durch die digitale Transformation hervorgerufenen gravierenden Veränderungen
wurden für Schule und Lehrerbildung bereits an verschiedenen Stellen aufgegriffen. So wurden
u.a. Kompetenzrahmen zur Beschreibung der Anforderungen an Lehrkräfte sowie zur praktischen
Orientierung geschaffen, u.a. das „TPACK-“ und das darauf aufbauende „DPACK-Modell“. Die
reflektierte Interaktion mit den mittlerweile breit verfügbaren Phänomenen der „KI-Welt“ erfordert
allerdings andere Kompetenzen als der Umgang mit herkömmlichen Informatiksystemen, weil „KISoftware“
auf einem anderen Gestaltungsansatz beruht, der dieser Software verschiedene spezielle
Eigenschaften verleiht. Dies stellt neue Anforderungen an die Professionalisierung von Lehrkräften
aller Fächer, weil die Entwicklung deutliche Auswirkungen auf die zentralen Handlungsfelder hat. Ziel
dieses Beitrages ist es, einen Rahmen zu definieren, der diese neuen KI-bezogenen Anforderungen
umreißt. Die vorliegende Arbeit nähert sich dem Thema theoretisch: Zunächst wird die Notwendigkeit
der Abgrenzung eines Bereichs der „KI-bezogenen Digitalitätskompetenz“ im DPACK-Modell
begründet. Da Lehrkräfte bei ihrer Tätigkeit pädagogische, fachliche und digitalisierungsbezogene
Kompetenzen kombinieren, bestimmen wir deduktiv die Bereiche der „KI-bezogenen pädagogischen
Kompetenz“ (AI-PK), der „KI-bezogenen fachinhaltlichen Kompetenz“ (AI-CK) und der „KIbezogenen
pädagogischen Inhaltskompetenz“ (AI-PCK) und illustrieren sie mit beispielhaften
Kompetenzzielen. Mit dem „AI-PACK“-Rahmen wird eine strukturierte Darstellung und Beforschung
der Anforderungen an KI-Bildung für Lehrkräfte ermöglicht.},
	booktitle = {Lecture {Notes} in {Informatics} ({LNI})},
	publisher = {Gesellschaft für Informatik},
	author = {Lorenz, Uwe and Romeike, Ralf},
	year = {2023},
	pages = {11--24},
}

Die durch die digitale Transformation hervorgerufenen gravierenden Veränderungen wurden für Schule und Lehrerbildung bereits an verschiedenen Stellen aufgegriffen. So wurden u.a. Kompetenzrahmen zur Beschreibung der Anforderungen an Lehrkräfte sowie zur praktischen Orientierung geschaffen, u.a. das „TPACK-“ und das darauf aufbauende „DPACK-Modell“. Die reflektierte Interaktion mit den mittlerweile breit verfügbaren Phänomenen der „KI-Welt“ erfordert allerdings andere Kompetenzen als der Umgang mit herkömmlichen Informatiksystemen, weil „KISoftware“ auf einem anderen Gestaltungsansatz beruht, der dieser Software verschiedene spezielle Eigenschaften verleiht. Dies stellt neue Anforderungen an die Professionalisierung von Lehrkräften aller Fächer, weil die Entwicklung deutliche Auswirkungen auf die zentralen Handlungsfelder hat. Ziel dieses Beitrages ist es, einen Rahmen zu definieren, der diese neuen KI-bezogenen Anforderungen umreißt. Die vorliegende Arbeit nähert sich dem Thema theoretisch: Zunächst wird die Notwendigkeit der Abgrenzung eines Bereichs der „KI-bezogenen Digitalitätskompetenz“ im DPACK-Modell begründet. Da Lehrkräfte bei ihrer Tätigkeit pädagogische, fachliche und digitalisierungsbezogene Kompetenzen kombinieren, bestimmen wir deduktiv die Bereiche der „KI-bezogenen pädagogischen Kompetenz“ (AI-PK), der „KI-bezogenen fachinhaltlichen Kompetenz“ (AI-CK) und der „KIbezogenen pädagogischen Inhaltskompetenz“ (AI-PCK) und illustrieren sie mit beispielhaften Kompetenzzielen. Mit dem „AI-PACK“-Rahmen wird eine strukturierte Darstellung und Beforschung der Anforderungen an KI-Bildung für Lehrkräfte ermöglicht.

@inproceedings{beersmans_training_2023,
	address = {Varna, Bulgaria},
	title = {Training and {Evaluation} of {Named} {Entity} {Recognition} {Models} for {Classical} {Latin}},
	url = {https://aclanthology.org/2023.alp-1.1.pdf},
	abstract = {We evaluate the performance of various models on the task of named entity recognition (NER) for classical Latin. Using an existing dataset, we train two transformer-based LatinBERT models and one shallow conditional random field (CRF) model. The performance is assessed using both standard metrics and a detailed manual error analysis, and compared to the results obtained by different already released Latin NER tools. Both analyses demonstrate that the BERT models achieve a better f1-score than the other models. Furthermore, we annotate new, unseen data for further evaluation of the models, and we discuss the impact of annotation choices on the results.},
	language = {English},
	booktitle = {Proceedings of the {Ancient} {Language} {Processing} {Workshop}},
	publisher = {INCOMA Ltd.},
	author = {Beersmans, Marijke and de Graaf, Evelien and Van de Cruys, Tim and Fantoli, Margherita},
	year = {2023},
	pages = {1--12},
}

We evaluate the performance of various models on the task of named entity recognition (NER) for classical Latin. Using an existing dataset, we train two transformer-based LatinBERT models and one shallow conditional random field (CRF) model. The performance is assessed using both standard metrics and a detailed manual error analysis, and compared to the results obtained by different already released Latin NER tools. Both analyses demonstrate that the BERT models achieve a better f1-score than the other models. Furthermore, we annotate new, unseen data for further evaluation of the models, and we discuss the impact of annotation choices on the results.

@inproceedings{berti_named_2023,
	address = {Graz},
	title = {Named {Entity} {Recognition} for a {Text}-{Based} {Catalog} of {Ancient} {Greek} {Authors} and {Works}},
	copyright = {Creative Commons Attribution 4.0 International, Open Access},
	url = {https://zenodo.org/record/8108058},
	doi = {10.5281/ZENODO.8108058},
	abstract = {This poster proposal presents a project whose results are the linguistic annotation of ancient Greek bibliographic references with a focus on Named Entity Recognition related to author names and work titles, in order to produce new dynamic text-based tools that are not available in existing indices and catalogs.},
	language = {en},
	urldate = {2025-01-26},
	booktitle = {Digital {Humanities} 2023: {Book} of {Abstracts}},
	publisher = {Zenodo},
	author = {Berti, Monica},
	year = {2023},
	keywords = {CITE Architecture, FOS: Languages and literature, Library \& information science, Linguistics, Literary studies, Paper, Philology, Poster, analysis, ancient Greek, and methods, concordancing and indexing, digital classics, digital philology, linked (open) data, literary canon, natural language processing, scholarly editing and editions development},
	pages = {557},
}

This poster proposal presents a project whose results are the linguistic annotation of ancient Greek bibliographic references with a focus on Named Entity Recognition related to author names and work titles, in order to produce new dynamic text-based tools that are not available in existing indices and catalogs.

@incollection{lanquillon_kunstliche_2023,
	address = {Wiesbaden},
	title = {Künstliche {Intelligenz}: {Ein} Überblick},
	isbn = {978-3-658-41689-8},
	shorttitle = {Künstliche {Intelligenz}},
	url = {https://doi.org/10.1007/978-3-658-41689-8_2},
	doi = {10.1007/978-3-658-41689-8_2},
	abstract = {Wir haben den Begriff Künstliche Intelligenz (KI) bereits mehrfach verwendet, ohne ihn explizit zu definieren. Was genau ist Künstliche Intelligenz? Gibt es eine exakte und allgemeingültige Definition? Seit wann gibt es den Begriff und wie hat sich das Thema als Fachgebiet über die Zeit entwickelt? Die folgenden Abschnitte sollen diese Fragen beantworten. Anschließend nehmen wir eine eher technische Perspektive ein, stellen Teilbereiche der KI vor und ordnen diese sie ein.},
	language = {de},
	urldate = {2024-09-05},
	booktitle = {Knowledge {Science} – {Grundlagen}: {Methoden} der {Künstlichen} {Intelligenz} für die {Wissensextraktion} aus {Texten}},
	publisher = {Springer Fachmedien},
	author = {Lanquillon, Carsten and Schacht, Sigurd},
	editor = {Lanquillon, Carsten and Schacht, Sigurd},
	year = {2023},
	pages = {5--21},
}

Wir haben den Begriff Künstliche Intelligenz (KI) bereits mehrfach verwendet, ohne ihn explizit zu definieren. Was genau ist Künstliche Intelligenz? Gibt es eine exakte und allgemeingültige Definition? Seit wann gibt es den Begriff und wie hat sich das Thema als Fachgebiet über die Zeit entwickelt? Die folgenden Abschnitte sollen diese Fragen beantworten. Anschließend nehmen wir eine eher technische Perspektive ein, stellen Teilbereiche der KI vor und ordnen diese sie ein.

@misc{beyer_data_2023,
	address = {Tübingen, Germany},
	title = {Data {Literacy} für die {Klassische} {Philologie} - {dAIdalos} - eine interaktive {Infrastruktur} als {Lernangebot}},
	url = {https://zenodo.org/record/8392485},
	doi = {10.5281/zenodo.8392485},
	abstract = {Das Poster informiert über das DFG-geförderte explorative Entwicklungsvorhaben Daidalos, das es Forschenden der Klassischen Philologie und verwandter Disziplinen ermöglichen soll, verschiedene Methoden des Natural Language Processing (NLP) an selbst zusammengestellten Forschungskorpora anzuwenden. Dabei ist Daidalos als interaktive Forschungsinfrastruktur konzipiert, die zugleich den Ausbau wesentlicher Teilfähigkeiten von Data Literacy, z. B. die Zusammenstellung und Analyse von Korpora oder den Umgang mit Annotationen, TEI-XML und graphischen Auswertungen, unterstützt. Hierzu sind vor allem forschungsorientierte, didaktische Lernbausteine und deren Implementierung in die Infrastruktur angedacht, um ein fach- und forschungsbezogenes Lernen zu ermöglichen.},
	urldate = {2023-10-09},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2023},
	keywords = {Data Literacy, FORGE2023, JupyterLab, KI-Didaktik, NLP, other},
}

Das Poster informiert über das DFG-geförderte explorative Entwicklungsvorhaben Daidalos, das es Forschenden der Klassischen Philologie und verwandter Disziplinen ermöglichen soll, verschiedene Methoden des Natural Language Processing (NLP) an selbst zusammengestellten Forschungskorpora anzuwenden. Dabei ist Daidalos als interaktive Forschungsinfrastruktur konzipiert, die zugleich den Ausbau wesentlicher Teilfähigkeiten von Data Literacy, z. B. die Zusammenstellung und Analyse von Korpora oder den Umgang mit Annotationen, TEI-XML und graphischen Auswertungen, unterstützt. Hierzu sind vor allem forschungsorientierte, didaktische Lernbausteine und deren Implementierung in die Infrastruktur angedacht, um ein fach- und forschungsbezogenes Lernen zu ermöglichen.

@misc{beyer_daidalos_2023,
	title = {Daidalos: {Forschen} und {Lernen} zugleich? {Data} {Literacy} als {Lernaufgabe} für die {Klassisch}-philologische {Forschung}},
	shorttitle = {Daidalos},
	url = {https://zenodo.org/record/8388900},
	doi = {10.5281/zenodo.8388900},
	abstract = {Poster zur Präsentation des Daidalos-Projekts beim "Workshop KI-Bildung. Ein Workshop zu Aus- und Weiterbildung über Künstliche Intelligenz im Rahmen der GI-Tagung INFORMATIK 2023" an der HTW Berlin},
	language = {deu},
	urldate = {2023-09-29},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = sep,
	year = {2023},
	keywords = {Digital Classics, data literacy, natural language processing, other, research infrastructure},
}

Poster zur Präsentation des Daidalos-Projekts beim "Workshop KI-Bildung. Ein Workshop zu Aus- und Weiterbildung über Künstliche Intelligenz im Rahmen der GI-Tagung INFORMATIK 2023" an der HTW Berlin

@phdthesis{yousef_translation_2023,
	address = {Leipzig},
	title = {Translation {Alignment} {Applied} to {Historical} {Languages}: methods, evaluation, applications, and visualization},
	shorttitle = {Translation {Alignment} {Applied} to {Historical} {Languages}},
	url = {https://nbn-resolving.org/urn:nbn:de:bsz:15-qucosa2-864719},
	school = {Universität Leipzig},
	author = {Yousef, Tariq},
	year = {2023},
	note = {https://ul.qucosa.de/api/qucosa\%3A86471/attachment/ATT-0/},
}

@inproceedings{yousef_transformer-based_2023,
	address = {Graz},
	title = {Transformer-{Based} {Named} {Entity} {Recognition} for {Ancient} {Greek}},
	url = {https://zenodo.org/records/8107629},
	abstract = {This paper presents our work on training two automatic NER models for ancient Greek using transformer-based models. The models classify the entities into three categories, namely, Person, Location, and Miscellaneous and achieved promising results on test and evaluation datasets.},
	language = {eng},
	urldate = {2024-01-04},
	booktitle = {Digital {Humanities} 2023: {Book} of {Abstracts}},
	author = {Yousef, Tariq and Palladino, Chiara and Janicke, Stefan},
	year = {2023},
	keywords = {Ancient Greek, Computer science, Cultural studies, Humanities computing, Linguistics, Long Presentation, Named Entities Recognition, Paper, Transformer models, and methods, annotation structures, natural language processing, systems},
	pages = {420--422},
}

This paper presents our work on training two automatic NER models for ancient Greek using transformer-based models. The models classify the entities into three categories, namely, Person, Location, and Miscellaneous and achieved promising results on test and evaluation datasets.

@inproceedings{yousef_classical_2023,
	address = {Varna, Bulgaria},
	title = {Classical {Philology} in the {Time} of {AI}: {Exploring} the {Potential} of {Parallel} {Corpora} in {Ancient} {Language}},
	shorttitle = {Classical {Philology} in the {Time} of {AI}},
	url = {https://aclanthology.org/2023.alp-1.21.pdf},
	abstract = {This paper provides an overview of diverse applications of parallel corpora in ancient languages, particularly Ancient Greek. In the first part, we provide the fundamental principles of parallel corpora and a short overview of their applications in the study of ancient texts. In the second part, we illustrate how to leverage on parallel corpora to perform various NLP tasks, including automatic translation alignment, dynamic lexica induction, and Named Entity Recognition. In the conclusions, we emphasize current limitations and future work.},
	urldate = {2025-01-26},
	booktitle = {Proceedings of the {Ancient} {Language} {Processing} {Workshop} associated with {RANLP}-2023},
	author = {Yousef, Tariq and Palladino, Chiara and Shamsian, Farnoosh},
	year = {2023},
	pages = {179--192},
}

This paper provides an overview of diverse applications of parallel corpora in ancient languages, particularly Ancient Greek. In the first part, we provide the fundamental principles of parallel corpora and a short overview of their applications in the study of ancient texts. In the second part, we illustrate how to leverage on parallel corpora to perform various NLP tasks, including automatic translation alignment, dynamic lexica induction, and Named Entity Recognition. In the conclusions, we emphasize current limitations and future work.

@article{laupichler_evaluating_2023,
	title = {Evaluating {AI} {Courses}: {A} {Valid} and {Reliable} {Instrument} for {Assessing} {Artificial}-{Intelligence} {Learning} through {Comparative} {Self}-{Assessment}},
	volume = {13},
	copyright = {http://creativecommons.org/licenses/by/3.0/},
	issn = {2227-7102},
	shorttitle = {Evaluating {AI} {Courses}},
	url = {https://www.mdpi.com/2227-7102/13/10/978},
	doi = {10.3390/educsci13100978},
	abstract = {A growing number of courses seek to increase the basic artificial-intelligence skills (“AI literacy”) of their participants. At this time, there is no valid and reliable measurement tool that can be used to assess AI-learning gains. However, the existence of such a tool would be important to enable quality assurance and comparability. In this study, a validated AI-literacy-assessment instrument, the “scale for the assessment of non-experts’ AI literacy” (SNAIL) was adapted and used to evaluate an undergraduate AI course. We investigated whether the scale can be used to reliably evaluate AI courses and whether mediator variables, such as attitudes toward AI or participation in other AI courses, had an influence on learning gains. In addition to the traditional mean comparisons (i.e., t-tests), the comparative self-assessment (CSA) gain was calculated, which allowed for a more meaningful assessment of the increase in AI literacy. We found preliminary evidence that the adapted SNAIL questionnaire enables a valid evaluation of AI-learning gains. In particular, distinctions among different subconstructs and the differentiation constructs, such as attitudes toward AI, seem to be possible with the help of the SNAIL questionnaire.},
	language = {en},
	number = {10},
	urldate = {2025-01-21},
	journal = {Education Sciences},
	publisher = {Multidisciplinary Digital Publishing Institute},
	author = {Laupichler, Matthias Carl and Aster, Alexandra and Perschewski, Jan-Ole and Schleiss, Johannes},
	month = oct,
	year = {2023},
	note = {Number: 10},
	keywords = {AI literacy, AI-literacy scale, artificial intelligence education, assessment, comparative self-assessment, course evaluation},
	pages = {978},
}

A growing number of courses seek to increase the basic artificial-intelligence skills (“AI literacy”) of their participants. At this time, there is no valid and reliable measurement tool that can be used to assess AI-learning gains. However, the existence of such a tool would be important to enable quality assurance and comparability. In this study, a validated AI-literacy-assessment instrument, the “scale for the assessment of non-experts’ AI literacy” (SNAIL) was adapted and used to evaluate an undergraduate AI course. We investigated whether the scale can be used to reliably evaluate AI courses and whether mediator variables, such as attitudes toward AI or participation in other AI courses, had an influence on learning gains. In addition to the traditional mean comparisons (i.e., t-tests), the comparative self-assessment (CSA) gain was calculated, which allowed for a more meaningful assessment of the increase in AI literacy. We found preliminary evidence that the adapted SNAIL questionnaire enables a valid evaluation of AI-learning gains. In particular, distinctions among different subconstructs and the differentiation constructs, such as attitudes toward AI, seem to be possible with the help of the SNAIL questionnaire.

@misc{xi_rise_2023,
	title = {The {Rise} and {Potential} of {Large} {Language} {Model} {Based} {Agents}: {A} {Survey}},
	shorttitle = {The {Rise} and {Potential} of {Large} {Language} {Model} {Based} {Agents}},
	url = {http://arxiv.org/abs/2309.07864},
	doi = {10.48550/arXiv.2309.07864},
	abstract = {For a long time, humanity has pursued artificial intelligence (AI) equivalent to or surpassing the human level, with AI agents considered a promising vehicle for this pursuit. AI agents are artificial entities that sense their environment, make decisions, and take actions. Many efforts have been made to develop intelligent agents, but they mainly focus on advancement in algorithms or training strategies to enhance specific capabilities or performance on particular tasks. Actually, what the community lacks is a general and powerful model to serve as a starting point for designing AI agents that can adapt to diverse scenarios. Due to the versatile capabilities they demonstrate, large language models (LLMs) are regarded as potential sparks for Artificial General Intelligence (AGI), offering hope for building general AI agents. Many researchers have leveraged LLMs as the foundation to build AI agents and have achieved significant progress. In this paper, we perform a comprehensive survey on LLM-based agents. We start by tracing the concept of agents from its philosophical origins to its development in AI, and explain why LLMs are suitable foundations for agents. Building upon this, we present a general framework for LLM-based agents, comprising three main components: brain, perception, and action, and the framework can be tailored for different applications. Subsequently, we explore the extensive applications of LLM-based agents in three aspects: single-agent scenarios, multi-agent scenarios, and human-agent cooperation. Following this, we delve into agent societies, exploring the behavior and personality of LLM-based agents, the social phenomena that emerge from an agent society, and the insights they offer for human society. Finally, we discuss several key topics and open problems within the field. A repository for the related papers at https://github.com/WooooDyy/LLM-Agent-Paper-List.},
	urldate = {2025-01-15},
	publisher = {arXiv},
	author = {Xi, Zhiheng and Chen, Wenxiang and Guo, Xin and He, Wei and Ding, Yiwen and Hong, Boyang and Zhang, Ming and Wang, Junzhe and Jin, Senjie and Zhou, Enyu and Zheng, Rui and Fan, Xiaoran and Wang, Xiao and Xiong, Limao and Zhou, Yuhao and Wang, Weiran and Jiang, Changhao and Zou, Yicheng and Liu, Xiangyang and Yin, Zhangyue and Dou, Shihan and Weng, Rongxiang and Cheng, Wensen and Zhang, Qi and Qin, Wenjuan and Zheng, Yongyan and Qiu, Xipeng and Huang, Xuanjing and Gui, Tao},
	month = sep,
	year = {2023},
	note = {arXiv:2309.07864 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language},
}

For a long time, humanity has pursued artificial intelligence (AI) equivalent to or surpassing the human level, with AI agents considered a promising vehicle for this pursuit. AI agents are artificial entities that sense their environment, make decisions, and take actions. Many efforts have been made to develop intelligent agents, but they mainly focus on advancement in algorithms or training strategies to enhance specific capabilities or performance on particular tasks. Actually, what the community lacks is a general and powerful model to serve as a starting point for designing AI agents that can adapt to diverse scenarios. Due to the versatile capabilities they demonstrate, large language models (LLMs) are regarded as potential sparks for Artificial General Intelligence (AGI), offering hope for building general AI agents. Many researchers have leveraged LLMs as the foundation to build AI agents and have achieved significant progress. In this paper, we perform a comprehensive survey on LLM-based agents. We start by tracing the concept of agents from its philosophical origins to its development in AI, and explain why LLMs are suitable foundations for agents. Building upon this, we present a general framework for LLM-based agents, comprising three main components: brain, perception, and action, and the framework can be tailored for different applications. Subsequently, we explore the extensive applications of LLM-based agents in three aspects: single-agent scenarios, multi-agent scenarios, and human-agent cooperation. Following this, we delve into agent societies, exploring the behavior and personality of LLM-based agents, the social phenomena that emerge from an agent society, and the insights they offer for human society. Finally, we discuss several key topics and open problems within the field. A repository for the related papers at https://github.com/WooooDyy/LLM-Agent-Paper-List.

@article{yousef_named_2023,
	title = {Named {Entity} {Annotation} {Projection} {Applied} to {Classical} {Languages}},
	url = {https://aclanthology.org/2023.latechclfl-1.19.pdf},
	abstract = {In this study, we demonstrate how to apply cross-lingual annotation projection to transfer named-entity annotations to classical languages for which limited or no resources and annotated texts are available, aiming to enrich their NER training datasets and train a model to perform NER tagging. Our approach employs sentence-level aligned corpora of ancient texts and the translation in a modern language, for which high-quality off-the-shelf NER systems are available. We automatically annotate the text of the modern language and employ a stateof-the-art neural word alignment system to find translation equivalents. Finally, we transfer the annotations to the corresponding tokens in the ancient texts using a direct projection heuristic. We applied our method to ancient Greek and Latin using the Bible with the English translation as a parallel corpus. We used the resulting annotations to enhance the performance of an existing NER model for ancient Greek.},
	language = {en},
	journal = {Proceedings of the 7th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature},
	author = {Yousef, Tariq and Palladino, Chiara and Heyer, Gerhard and Jänicke, Stefan},
	year = {2023},
	pages = {175--182},
}

In this study, we demonstrate how to apply cross-lingual annotation projection to transfer named-entity annotations to classical languages for which limited or no resources and annotated texts are available, aiming to enrich their NER training datasets and train a model to perform NER tagging. Our approach employs sentence-level aligned corpora of ancient texts and the translation in a modern language, for which high-quality off-the-shelf NER systems are available. We automatically annotate the text of the modern language and employ a stateof-the-art neural word alignment system to find translation equivalents. Finally, we transfer the annotations to the corresponding tokens in the ancient texts using a direct projection heuristic. We applied our method to ancient Greek and Latin using the Bible with the English translation as a parallel corpus. We used the resulting annotations to enhance the performance of an existing NER model for ancient Greek.

@article{rebora_sentiment_2023,
	title = {Sentiment {Analysis} in {Literary} {Studies}. {A} {Critical} {Survey}},
	volume = {17},
	url = {https://digitalhumanities.org/dhq/vol/17/2/000691/000691.html},
	number = {2},
	journal = {digital humanities quaterly},
	author = {Rebora, Simone},
	year = {2023},
}

@article{sprugnoli_sentiment_2023,
	title = {The {Sentiment} of {Latin} {Poetry}. {Annotation} and {Automatic} {Analysis} of the {Odes} of {Horace}},
	volume = {9},
	issn = {2499-4553},
	url = {https://journals.openedition.org/ijcol/1125},
	number = {9-1},
	journal = {IJCoL. Italian Journal of Computational Linguistics},
	publisher = {Accademia University Press},
	author = {Sprugnoli, Rachele and Mambrini, Francesco and Passarotti, Marco and Moretti, Giovanni},
	year = {2023},
	pages = {53--71},
}

@article{hambarde_information_2023,
	title = {Information {Retrieval}: {Recent} {Advances} and {Beyond}},
	volume = {11},
	issn = {2169-3536},
	shorttitle = {Information {Retrieval}},
	url = {http://arxiv.org/abs/2301.08801},
	doi = {10.1109/ACCESS.2023.3295776},
	abstract = {In this paper, we provide a detailed overview of the models used for information retrieval in the first and second stages of the typical processing chain. We discuss the current state-of-the-art models, including methods based on terms, semantic retrieval, and neural. Additionally, we delve into the key topics related to the learning process of these models. This way, this survey offers a comprehensive understanding of the field and is of interest for for researchers and practitioners entering/working in the information retrieval domain.},
	urldate = {2024-09-03},
	journal = {IEEE Access},
	author = {Hambarde, Kailash A. and Proenca, Hugo},
	year = {2023},
	note = {arXiv:2301.08801 [cs]},
	keywords = {Computer Science - Information Retrieval},
	pages = {76581--76604},
}

In this paper, we provide a detailed overview of the models used for information retrieval in the first and second stages of the typical processing chain. We discuss the current state-of-the-art models, including methods based on terms, semantic retrieval, and neural. Additionally, we delve into the key topics related to the learning process of these models. This way, this survey offers a comprehensive understanding of the field and is of interest for for researchers and practitioners entering/working in the information retrieval domain.

@inproceedings{celano_neural_2023,
	title = {A {Neural} {Network} {Approach} to {Ellipsis} {Detection} in {Ancient} {Greek}},
	author = {Celano, Giuseppe GA},
	year = {2023},
	pages = {151--158},
}

@article{ross_new_2023,
	title = {A {New} {Frontier}: {AI} and {Ancient} {Language} {Pedagogy}},
	volume = {24},
	issn = {2058-6310},
	shorttitle = {A {New} {Frontier}},
	url = {https://www.cambridge.org/core/journals/journal-of-classics-teaching/article/new-frontier-ai-and-ancient-language-pedagogy/A63EF69F5FE5529F0F45FB1EB655A9F7},
	doi = {10.1017/S2058631023000430},
	abstract = {In November 2022, ChatGPT 3.5 was released on a public research preview, gaining notoriety for its ability to pull from a vast body of information to create coherent and digestible bodies of text that accurately respond to queries (OpenAI, 2022). It is able to recognise the grammar and vocabulary of ancient languages, translate passages, and compose texts at an alarmingly accurate and rapid rate. For teachers, this AI has had mixed reviews. Some fear its ability to produce well-written work effortlessly, while others are excited by its abilities to push the boundaries of current teaching practices. This paper explores how well ChatGPT explains grammatical concepts, parses inflected forms, and translates Classical Latin, Ancient Greek, and Classical Sanskrit. Overall, ChatGPT is rather good at working with Classical Latin and Sanskrit, but its abilities with Ancient Greek are deeply problematic. Although it is quite flawed at this time, ChatGPT, when used properly, could become a useful a tool for ancient language study. With proper guiding phrases, students could use this AI to practise vocabulary, check their translations, and rephrase grammatical concepts.},
	language = {en},
	number = {48},
	urldate = {2024-03-26},
	journal = {Journal of Classics Teaching},
	author = {Ross, Edward A. S.},
	month = oct,
	year = {2023},
	keywords = {Ancient Greek, Ancient Language Pedagogy, Artificial Intelligence, ChatGPT, Classical Latin, Classical Sanskrit, New Teaching Tools},
	pages = {143--161},
}

In November 2022, ChatGPT 3.5 was released on a public research preview, gaining notoriety for its ability to pull from a vast body of information to create coherent and digestible bodies of text that accurately respond to queries (OpenAI, 2022). It is able to recognise the grammar and vocabulary of ancient languages, translate passages, and compose texts at an alarmingly accurate and rapid rate. For teachers, this AI has had mixed reviews. Some fear its ability to produce well-written work effortlessly, while others are excited by its abilities to push the boundaries of current teaching practices. This paper explores how well ChatGPT explains grammatical concepts, parses inflected forms, and translates Classical Latin, Ancient Greek, and Classical Sanskrit. Overall, ChatGPT is rather good at working with Classical Latin and Sanskrit, but its abilities with Ancient Greek are deeply problematic. Although it is quite flawed at this time, ChatGPT, when used properly, could become a useful a tool for ancient language study. With proper guiding phrases, students could use this AI to practise vocabulary, check their translations, and rephrase grammatical concepts.

@article{krahn_sentence_2023,
	title = {Sentence embedding models for {Ancient} {Greek} using multilingual knowledge distillation},
	journal = {arXiv preprint arXiv:2308.13116},
	author = {Krahn, Kevin and Tate, Derrick and Lamicela, Andrew C},
	year = {2023},
}

@article{sansom_sedes_2023,
	title = {{SEDES}: {Metrical} {Position} in {Greek} {Hexameter}},
	volume = {017},
	issn = {1938-4122},
	shorttitle = {{SEDES}},
	url = {https://digitalhumanities.org/dhq/vol/17/2/000675/000675.html},
	number = {2},
	journal = {Digital Humanities Quarterly},
	author = {Sansom, Stephen A. and Fifield, David},
	month = may,
	year = {2023},
}

@article{stopponi_evaluation_2023,
	title = {Evaluation of {Distributional} {Semantic} {Models} of {Ancient} {Greek}: {Ancient} {Language} {Processing}},
	shorttitle = {Evaluation of {Distributional} {Semantic} {Models} of {Ancient} {Greek}},
	url = {https://pure.rug.nl/ws/portalfiles/portal/777728242/stopponi_et_al_2023_alp.pdf},
	abstract = {We evaluate four count-based and predictive distributional semantic models of Ancient Greek against AGREE, a composite benchmark of human judgements, to assess their ability to retrieve semantic relatedness. On the basis of the observations deriving from the analysis of the results, we design a procedure for a largerscale intrinsic evaluation of count-based and predictive language models, including syntactic embeddings. We also propose possible ways of exploiting the different layers of the whole AGREE benchmark (including both humanand machine-generated data) and different evaluation metrics.},
	journal = {Proceedings of the Ancient Language Processing Workshop},
	publisher = {Association for Computational Linguistics (ACL)},
	author = {Stopponi, Silvia and Pedrazzini, Nilo and Peels-Matthey, Saskia and McGillivray, Barbara and Nissim, Malvina},
	month = sep,
	year = {2023},
	note = {Place: Varna, Bulgaria},
	keywords = {ancient greek, ancient languages, benchmark, evaluation, languge models, natural language processing, word2vec},
	pages = {49--58},
}

We evaluate four count-based and predictive distributional semantic models of Ancient Greek against AGREE, a composite benchmark of human judgements, to assess their ability to retrieve semantic relatedness. On the basis of the observations deriving from the analysis of the results, we design a procedure for a largerscale intrinsic evaluation of count-based and predictive language models, including syntactic embeddings. We also propose possible ways of exploiting the different layers of the whole AGREE benchmark (including both humanand machine-generated data) and different evaluation metrics.

@misc{wang_gpt-ner_2023,
	title = {{GPT}-{NER}: {Named} {Entity} {Recognition} via {Large} {Language} {Models}},
	shorttitle = {{GPT}-{NER}},
	url = {http://arxiv.org/abs/2304.10428},
	doi = {10.48550/arXiv.2304.10428},
	abstract = {Despite the fact that large-scale Language Models (LLM) have achieved SOTA performances on a variety of NLP tasks, its performance on NER is still significantly below supervised baselines. This is due to the gap between the two tasks the NER and LLMs: the former is a sequence labeling task in nature while the latter is a text-generation model. In this paper, we propose GPT-NER to resolve this issue. GPT-NER bridges the gap by transforming the sequence labeling task to a generation task that can be easily adapted by LLMs e.g., the task of finding location entities in the input text "Columbus is a city" is transformed to generate the text sequence "@@Columbus\#\# is a city", where special tokens @@\#\# marks the entity to extract. To efficiently address the "hallucination" issue of LLMs, where LLMs have a strong inclination to over-confidently label NULL inputs as entities, we propose a self-verification strategy by prompting LLMs to ask itself whether the extracted entities belong to a labeled entity tag. We conduct experiments on five widely adopted NER datasets, and GPT-NER achieves comparable performances to fully supervised baselines, which is the first time as far as we are concerned. More importantly, we find that GPT-NER exhibits a greater ability in the low-resource and few-shot setups, when the amount of training data is extremely scarce, GPT-NER performs significantly better than supervised models. This demonstrates the capabilities of GPT-NER in real-world NER applications where the number of labeled examples is limited.},
	urldate = {2024-06-21},
	publisher = {arXiv},
	author = {Wang, Shuhe and Sun, Xiaofei and Li, Xiaoya and Ouyang, Rongbin and Wu, Fei and Zhang, Tianwei and Li, Jiwei and Wang, Guoyin},
	month = oct,
	year = {2023},
	note = {arXiv:2304.10428 [cs]},
	keywords = {Computer Science - Computation and Language},
}

Despite the fact that large-scale Language Models (LLM) have achieved SOTA performances on a variety of NLP tasks, its performance on NER is still significantly below supervised baselines. This is due to the gap between the two tasks the NER and LLMs: the former is a sequence labeling task in nature while the latter is a text-generation model. In this paper, we propose GPT-NER to resolve this issue. GPT-NER bridges the gap by transforming the sequence labeling task to a generation task that can be easily adapted by LLMs e.g., the task of finding location entities in the input text "Columbus is a city" is transformed to generate the text sequence "@@Columbus## is a city", where special tokens @@## marks the entity to extract. To efficiently address the "hallucination" issue of LLMs, where LLMs have a strong inclination to over-confidently label NULL inputs as entities, we propose a self-verification strategy by prompting LLMs to ask itself whether the extracted entities belong to a labeled entity tag. We conduct experiments on five widely adopted NER datasets, and GPT-NER achieves comparable performances to fully supervised baselines, which is the first time as far as we are concerned. More importantly, we find that GPT-NER exhibits a greater ability in the low-resource and few-shot setups, when the amount of training data is extremely scarce, GPT-NER performs significantly better than supervised models. This demonstrates the capabilities of GPT-NER in real-world NER applications where the number of labeled examples is limited.

@misc{myerston_grecy_2023,
	title = {{greCy}: {Ancient} {Greek} {spaCy} models for {Natural} {Language} {Processing} in {Python}},
	copyright = {MIT},
	shorttitle = {{greCy}},
	url = {https://github.com/jmyerston/greCy},
	abstract = {Ancient Greek language models for spaCy},
	urldate = {2024-02-27},
	author = {Myerston, Jacobo and López, Jose},
	month = dec,
	year = {2023},
	note = {original-date: 2022-09-18T23:13:41Z},
}

Ancient Greek language models for spaCy

@article{sommerschield_machine_2023,
	title = {Machine {Learning} for {Ancient} {Languages}: {A} {Survey}},
	issn = {0891-2017},
	shorttitle = {Machine {Learning} for {Ancient} {Languages}},
	url = {https://doi.org/10.1162/coli_a_00481},
	doi = {10.1162/coli_a_00481},
	abstract = {Ancient languages preserve the cultures and histories of the past. However, their study is fraught with difficulties, and experts must tackle a range of challenging text-based tasks, from deciphering lost languages to restoring damaged inscriptions, to determining the authorship of works of literature. Technological aids have long supported the study of ancient texts, but in recent years advances in artificial intelligence and machine learning have enabled analyses on a scale and in a detail that are reshaping the field of humanities, similarly to how microscopes and telescopes have contributed to the realm of science. This article aims to provide a comprehensive survey of published research using machine learning for the study of ancient texts written in any language, script, and medium, spanning over three and a half millennia of civilizations around the ancient world. To analyze the relevant literature, we introduce a taxonomy of tasks inspired by the steps involved in the study of ancient documents: digitization, restoration, attribution, linguistic analysis, textual criticism, translation, and decipherment. This work offers three major contributions: first, mapping the interdisciplinary field carved out by the synergy between the humanities and machine learning; second, highlighting how active collaboration between specialists from both fields is key to producing impactful and compelling scholarship; third, highlighting promising directions for future work in this field. Thus, this work promotes and supports the continued collaborative impetus between the humanities and machine learning.},
	urldate = {2023-09-15},
	journal = {Computational Linguistics},
	author = {Sommerschield, Thea and Assael, Yannis and Pavlopoulos, John and Stefanak, Vanessa and Senior, Andrew and Dyer, Chris and Bodel, John and Prag, Jonathan and Androutsopoulos, Ion and Freitas, Nando de},
	month = aug,
	year = {2023},
	pages = {1--45},
}

Ancient languages preserve the cultures and histories of the past. However, their study is fraught with difficulties, and experts must tackle a range of challenging text-based tasks, from deciphering lost languages to restoring damaged inscriptions, to determining the authorship of works of literature. Technological aids have long supported the study of ancient texts, but in recent years advances in artificial intelligence and machine learning have enabled analyses on a scale and in a detail that are reshaping the field of humanities, similarly to how microscopes and telescopes have contributed to the realm of science. This article aims to provide a comprehensive survey of published research using machine learning for the study of ancient texts written in any language, script, and medium, spanning over three and a half millennia of civilizations around the ancient world. To analyze the relevant literature, we introduce a taxonomy of tasks inspired by the steps involved in the study of ancient documents: digitization, restoration, attribution, linguistic analysis, textual criticism, translation, and decipherment. This work offers three major contributions: first, mapping the interdisciplinary field carved out by the synergy between the humanities and machine learning; second, highlighting how active collaboration between specialists from both fields is key to producing impactful and compelling scholarship; third, highlighting promising directions for future work in this field. Thus, this work promotes and supports the continued collaborative impetus between the humanities and machine learning.

@article{du_shortcut_2023,
	title = {Shortcut {Learning} of {Large} {Language} {Models} in {Natural} {Language} {Understanding}},
	volume = {67},
	issn = {0001-0782},
	url = {https://dl.acm.org/doi/10.1145/3596490},
	doi = {10.1145/3596490},
	abstract = {Shortcuts often hinder the robustness of large language models.},
	number = {1},
	urldate = {2024-05-01},
	journal = {Communications of the ACM},
	author = {Du, Mengnan and He, Fengxiang and Zou, Na and Tao, Dacheng and Hu, Xia},
	year = {2023},
	pages = {110--120},
}

Shortcuts often hinder the robustness of large language models.

@misc{riemenschneider_exploring_2023,
	title = {Exploring {Large} {Language} {Models} for {Classical} {Philology}},
	url = {http://arxiv.org/abs/2305.13698},
	doi = {10.48550/arXiv.2305.13698},
	abstract = {Recent advances in NLP have led to the creation of powerful language models for many languages including Ancient Greek and Latin. While prior work on Classical languages unanimously uses BERT, in this work we create four language models for Ancient Greek that vary along two dimensions to study their versatility for tasks of interest for Classical languages: we explore (i) encoder-only and encoder-decoder architectures using RoBERTa and T5 as strong model types, and create for each of them (ii) a monolingual Ancient Greek and a multilingual instance that includes Latin and English. We evaluate all models on morphological and syntactic tasks, including lemmatization, which demonstrates the added value of T5's decoding abilities. We further define two probing tasks to investigate the knowledge acquired by models pre-trained on Classical texts. Our experiments provide the first benchmarking analysis of existing models of Ancient Greek. Results show that our models provide significant improvements over the SoTA. The systematic analysis of model types can inform future research in designing language models for Classical languages, including the development of novel generative tasks. We make all our models available as community resources, along with a large curated pre-training corpus for Ancient Greek, to support the creation of a larger, comparable model zoo for Classical Philology. Our models and resources are available at https://github.com/Heidelberg-NLP/ancient-language-models.},
	urldate = {2023-09-29},
	publisher = {arXiv},
	author = {Riemenschneider, Frederick and Frank, Anette},
	month = may,
	year = {2023},
	note = {arXiv:2305.13698 [cs]},
	keywords = {Computer Science - Computation and Language, I.2.7},
}

Recent advances in NLP have led to the creation of powerful language models for many languages including Ancient Greek and Latin. While prior work on Classical languages unanimously uses BERT, in this work we create four language models for Ancient Greek that vary along two dimensions to study their versatility for tasks of interest for Classical languages: we explore (i) encoder-only and encoder-decoder architectures using RoBERTa and T5 as strong model types, and create for each of them (ii) a monolingual Ancient Greek and a multilingual instance that includes Latin and English. We evaluate all models on morphological and syntactic tasks, including lemmatization, which demonstrates the added value of T5's decoding abilities. We further define two probing tasks to investigate the knowledge acquired by models pre-trained on Classical texts. Our experiments provide the first benchmarking analysis of existing models of Ancient Greek. Results show that our models provide significant improvements over the SoTA. The systematic analysis of model types can inform future research in designing language models for Classical languages, including the development of novel generative tasks. We make all our models available as community resources, along with a large curated pre-training corpus for Ancient Greek, to support the creation of a larger, comparable model zoo for Classical Philology. Our models and resources are available at https://github.com/Heidelberg-NLP/ancient-language-models.

@article{salden_didaktische_2023,
	title = {Didaktische und rechtliche {Perspektiven} auf {KI}-gestütztes {Schreiben} in der {Hochschulbildung}},
	journal = {Zentrum für Wissenschaftsdidaktik der Ruhr-Universität Bochum},
	author = {Salden, Peter and Leschke, Jonas},
	year = {2023},
}

@article{hartman_quantitative_2023,
	title = {{QUANTITATIVE} {APPROACHES} {TO} {LATE} {ANTIQUE} {POETICS}: {ENUMERATION} {AND} {CONGERIES}},
	issn = {1350346411},
	journal = {A Late Antique Poetics?: The Jeweled Style Revisited},
	publisher = {Bloomsbury Publishing},
	author = {Hartman, Joshua and Levernier, Jacob},
	year = {2023},
	pages = {75},
}

@inproceedings{warstadt_findings_2023,
	address = {Singapore},
	title = {Findings of the {BabyLM} {Challenge}: {Sample}-{Efficient} {Pretraining} on {Developmentally} {Plausible} {Corpora}},
	shorttitle = {Findings of the {BabyLM} {Challenge}},
	url = {https://aclanthology.org/2023.conll-babylm.1},
	doi = {10.18653/v1/2023.conll-babylm.1},
	abstract = {Children can acquire language from less than 100 million words of input. Large language models are far less data-efficient: they typically require 3 or 4 orders of magnitude more data and still do not perform as well as humans on many evaluations. These intensive resource demands limit the ability of researchers to train new models and use existing models as developmentally plausible cognitive models. The BabyLM Challenge is a communal effort in which participants compete to optimize language model training on a fixed data budget. Submissions are compared on various evaluation tasks targeting grammatical ability, downstream task performance, and generalization. Participants can submit to up to three tracks with progressively looser data restrictions. From over 30 submissions, we extract concrete recommendations on how best to train data-efficient language models, and on where future efforts should (and perhaps should not) focus. The winning submissions using the LTG-BERT architecture (Samuel et al., 2023) outperformed models trained on trillions of words. Other submissions achieved strong results through training on shorter input sequences or training a student model on a pretrained teacher. Curriculum learning attempts, which accounted for a large number of submissions, were largely unsuccessful, though some showed modest improvements.},
	language = {en},
	urldate = {2024-01-16},
	booktitle = {Proceedings of the {BabyLM} {Challenge} at the 27th {Conference} on {Computational} {Natural} {Language} {Learning}},
	publisher = {Association for Computational Linguistics},
	author = {Warstadt, Alex and Mueller, Aaron and Choshen, Leshem and Wilcox, Ethan and Zhuang, Chengxu and Ciro, Juan and Mosquera, Rafael and Paranjabe, Bhargavi and Williams, Adina and Linzen, Tal and Cotterell, Ryan},
	year = {2023},
	pages = {1--6},
}

Children can acquire language from less than 100 million words of input. Large language models are far less data-efficient: they typically require 3 or 4 orders of magnitude more data and still do not perform as well as humans on many evaluations. These intensive resource demands limit the ability of researchers to train new models and use existing models as developmentally plausible cognitive models. The BabyLM Challenge is a communal effort in which participants compete to optimize language model training on a fixed data budget. Submissions are compared on various evaluation tasks targeting grammatical ability, downstream task performance, and generalization. Participants can submit to up to three tracks with progressively looser data restrictions. From over 30 submissions, we extract concrete recommendations on how best to train data-efficient language models, and on where future efforts should (and perhaps should not) focus. The winning submissions using the LTG-BERT architecture (Samuel et al., 2023) outperformed models trained on trillions of words. Other submissions achieved strong results through training on shorter input sequences or training a student model on a pretrained teacher. Curriculum learning attempts, which accounted for a large number of submissions, were largely unsuccessful, though some showed modest improvements.

@inproceedings{kostkan_odycy_2023,
	address = {Dubrovnik, Croatia},
	title = {{OdyCy} – {A} general-purpose {NLP} pipeline for {Ancient} {Greek}},
	booktitle = {Proceedings of the 7th {Joint} {SIGHUM} {Workshop} on {Computational} {Linguistics} for {Cultural} {Heritage}, {Social} {Sciences}, {Humanities} and {Literature}},
	publisher = {Association for Computational Linguistics},
	author = {Kostkan, Jan and Kardos, Márton and Palle Bliddal Mortensen, Jacob and Laigaard Nielbo, Kristofer},
	year = {2023},
	pages = {128--134},
}

@incollection{gamba_latin_2023,
	address = {Varna, Bulgaria},
	title = {Latin {Morphology} through the {Centuries}: {Ensuring} {Consistency} for {Better} {Language} {Processing}.},
	url = {https://ufal.mff.cuni.cz/biblio/attachments/2023-gamba-p3787387064232511302.pdf},
	booktitle = {Proceedings of the {Ancient} {Language} {Processing} {Workshop} associated with the 14th {International} {Conference} on {Recent} {Advances} in {Natural} {Language} {Processing} {RANLP} 2023},
	author = {Gamba, F. and Zeman, D.},
	year = {2023},
}

@incollection{gamba_universalising_2023,
	address = {Washington, DC, USA},
	series = {Proceedings of the {Sixth} {Workshop} on {Universal} {Dependencies} ({UDW}, {GURT}/{SyntaxFest} 2023)},
	title = {Universalising {Latin} {Universal} {Dependencies}: a harmonisation of {Latin} treebanks in {UD}.},
	url = {https://aclanthology.org/2023.udw-1.2/},
	publisher = {March. Association for Computational Linguistics (ACL)},
	author = {Gamba, F. and Zeman, D.},
	year = {2023},
}

@article{grillo_meta-literature_2023,
	title = {Meta-{Literature} and {Mimesis} in the {Rhetorica} ad {Herennium}},
	volume = {144},
	number = {1},
	journal = {American Journal of Philology},
	author = {Grillo, Luca},
	year = {2023},
	pages = {41--72},
}

@misc{riemenschneider_graecia_2023,
	title = {Graecia capta ferum victorem cepit. {Detecting} {Latin} {Allusions} to {Ancient} {Greek} {Literature}},
	url = {http://arxiv.org/abs/2308.12008},
	doi = {10.48550/arXiv.2308.12008},
	abstract = {Intertextual allusions hold a pivotal role in Classical Philology, with Latin authors frequently referencing Ancient Greek texts. Until now, the automatic identification of these intertextual references has been constrained to monolingual approaches, seeking parallels solely within Latin or Greek texts. In this study, we introduce SPhilBERTa, a trilingual Sentence-RoBERTa model tailored for Classical Philology, which excels at cross-lingual semantic comprehension and identification of identical sentences across Ancient Greek, Latin, and English. We generate new training data by automatically translating English texts into Ancient Greek. Further, we present a case study, demonstrating SPhilBERTa's capability to facilitate automated detection of intertextual parallels. Our models and resources are available at https://github.com/Heidelberg-NLP/ancient-language-models.},
	urldate = {2023-09-29},
	publisher = {arXiv},
	author = {Riemenschneider, Frederick and Frank, Anette},
	month = aug,
	year = {2023},
	note = {arXiv:2308.12008 [cs]},
	keywords = {Computer Science - Computation and Language, I.2.7},
}

Intertextual allusions hold a pivotal role in Classical Philology, with Latin authors frequently referencing Ancient Greek texts. Until now, the automatic identification of these intertextual references has been constrained to monolingual approaches, seeking parallels solely within Latin or Greek texts. In this study, we introduce SPhilBERTa, a trilingual Sentence-RoBERTa model tailored for Classical Philology, which excels at cross-lingual semantic comprehension and identification of identical sentences across Ancient Greek, Latin, and English. We generate new training data by automatically translating English texts into Ancient Greek. Further, we present a case study, demonstrating SPhilBERTa's capability to facilitate automated detection of intertextual parallels. Our models and resources are available at https://github.com/Heidelberg-NLP/ancient-language-models.

@article{krahn_sentence_2023,
	title = {Sentence {Embedding} {Models} for {Ancient} {Greek} {Using} {Multilingual} {Knowledge} {Distillation}},
	url = {https://arxiv.org/pdf/2308.13116.pdf},
	journal = {arXiv preprint arXiv:2308.13116},
	author = {Krahn, Kevin and Tate, Derrick and Lamicela, Andrew C},
	year = {2023},
}

@article{van_der_lek_integrating_2023,
	title = {Integrating research infrastructures into teaching: {Recommendations} and best practices},
	shorttitle = {Integrating research infrastructures into teaching},
	url = {https://zenodo.org/record/8114407},
	abstract = {The UPSKILLS needs revealed that linguistics and language-related degree programmes seldom include language data standards and research data repositories in their learning outcomes. A survey of lecturers from linguistics and language-related disciplines also exposed a number of challenges in using repositories for language data discovery, reuse and archiving. Against this backdrop, the present guide shows how teachers and trainers can leverage the CLARIN research infrastructure to help students enhance their data collection, processing and analysis, and archiving skills. By integrating research infrastructures into teaching, educators can bridge the gap between theoretical knowledge and practical aspects of linguistic research data management, equipping students with the necessary skills and competences to thrive in the evolving landscape of open science and data-driven research.},
	urldate = {2023-10-02},
	publisher = {Zenodo},
	author = {van der Lek, Iulianna and Fišer, Darja and Samardzic, Tanja and Simonovic, Marko and Assimakopoulos, Stavros and Bernardini, Silvia and Milicevic Petrovic, Maja and Puskas, Genoveva},
	month = aug,
	year = {2023},
}

The UPSKILLS needs revealed that linguistics and language-related degree programmes seldom include language data standards and research data repositories in their learning outcomes. A survey of lecturers from linguistics and language-related disciplines also exposed a number of challenges in using repositories for language data discovery, reuse and archiving. Against this backdrop, the present guide shows how teachers and trainers can leverage the CLARIN research infrastructure to help students enhance their data collection, processing and analysis, and archiving skills. By integrating research infrastructures into teaching, educators can bridge the gap between theoretical knowledge and practical aspects of linguistic research data management, equipping students with the necessary skills and competences to thrive in the evolving landscape of open science and data-driven research.

@misc{schork_kunstliche_2023,
	title = {Künstliche {Intelligenz} in der {Bildung}: {Drei} {Zukunftsszenarien} und fünf {Handlungsfelder}},
	url = {https://ki-campus.org/sites/default/files/2023-04/2023-03_Diskussionspapier_KI_Bildung_Zukunftsszenarien_Handlungsfelder_KI-Campus.pdf},
	abstract = {Die voranschreitende Digitalisierung und insbesondere der Einsatz von Künstlicher Intelligenz (KI) in der
Bildung eröffnen neue Möglichkeiten des Lernens. Bildung kann stärker individuell sowie zeitlich und räumlich entgrenzt stattfinden. ChatGPT verdeutlicht, wie dynamisch die Entwicklungen im Bereich KI sind. Das in Zunahme begriffene Interesse am Thema KI allgemein sowie KI in der Bildung kommt dem erforderlichen Austausch zwischen zentralen Akteur:innen zugute. Dieses Diskussionspapier präsentiert drei Zukunftsszenarien für den Einsatz von KI in der institutionellen Bildung: Hochschule, Weiterbildung
und Schule. Im Fokus stehen die übergeordneten Fragestellungen nach Voraussetzungen und
Gelingensbedingungen einer erfolgreichen Anwendung sowie nach gemeinsamen Handlungsfeldern für zentrale Stakeholder. Fünf Handlungsfelder für den Einsatz von KI in institutioneller Bildung sind auf Grundlage der skizzierten Zukunftsszenarien hervorzuheben: (1) Interdisziplinäre Zusammenarbeit, (2)
Qualifizierungsangebote und Kompetenzentwicklung, (3) Digitale Infrastruktur und Personal, (4) Ethik und Datensouveränität sowie (5) Interoperabilität von Daten in Bildungskontexten. Ziel des
vorliegenden Diskussionspapiers ist es, den kooperativen Austausch zwischen relevanten Stakeholdern anzuregen.},
	author = {Schork, Sabrina and Schleiss, Johannes and Mah, Dana-Kristin and Böhme, Katrin and Fischer, David and Mesenhöller, Janne and Paaßen, Benjamin and Schrumpf, Johannes},
	year = {2023},
}

Die voranschreitende Digitalisierung und insbesondere der Einsatz von Künstlicher Intelligenz (KI) in der Bildung eröffnen neue Möglichkeiten des Lernens. Bildung kann stärker individuell sowie zeitlich und räumlich entgrenzt stattfinden. ChatGPT verdeutlicht, wie dynamisch die Entwicklungen im Bereich KI sind. Das in Zunahme begriffene Interesse am Thema KI allgemein sowie KI in der Bildung kommt dem erforderlichen Austausch zwischen zentralen Akteur:innen zugute. Dieses Diskussionspapier präsentiert drei Zukunftsszenarien für den Einsatz von KI in der institutionellen Bildung: Hochschule, Weiterbildung und Schule. Im Fokus stehen die übergeordneten Fragestellungen nach Voraussetzungen und Gelingensbedingungen einer erfolgreichen Anwendung sowie nach gemeinsamen Handlungsfeldern für zentrale Stakeholder. Fünf Handlungsfelder für den Einsatz von KI in institutioneller Bildung sind auf Grundlage der skizzierten Zukunftsszenarien hervorzuheben: (1) Interdisziplinäre Zusammenarbeit, (2) Qualifizierungsangebote und Kompetenzentwicklung, (3) Digitale Infrastruktur und Personal, (4) Ethik und Datensouveränität sowie (5) Interoperabilität von Daten in Bildungskontexten. Ziel des vorliegenden Diskussionspapiers ist es, den kooperativen Austausch zwischen relevanten Stakeholdern anzuregen.

@article{motoki_more_2023,
	title = {More human than human: measuring {ChatGPT} political bias},
	issn = {1573-7101},
	shorttitle = {More human than human},
	url = {https://doi.org/10.1007/s11127-023-01097-2},
	doi = {10.1007/s11127-023-01097-2},
	abstract = {We investigate the political bias of a large language model (LLM), ChatGPT, which has become popular for retrieving factual information and generating content. Although ChatGPT assures that it is impartial, the literature suggests that LLMs exhibit bias involving race, gender, religion, and political orientation. Political bias in LLMs can have adverse political and electoral consequences similar to bias from traditional and social media. Moreover, political bias can be harder to detect and eradicate than gender or racial bias. We propose a novel empirical design to infer whether ChatGPT has political biases by requesting it to impersonate someone from a given side of the political spectrum and comparing these answers with its default. We also propose dose-response, placebo, and profession-politics alignment robustness tests. To reduce concerns about the randomness of the generated text, we collect answers to the same questions 100 times, with question order randomized on each round. We find robust evidence that ChatGPT presents a significant and systematic political bias toward the Democrats in the US, Lula in Brazil, and the Labour Party in the UK. These results translate into real concerns that ChatGPT, and LLMs in general, can extend or even amplify the existing challenges involving political processes posed by the Internet and social media. Our findings have important implications for policymakers, media, politics, and academia stakeholders.},
	language = {en},
	urldate = {2023-09-15},
	journal = {Public Choice},
	author = {Motoki, Fabio and Neto, Valdemar Pinho and Rodrigues, Victor},
	month = aug,
	year = {2023},
	keywords = {Bias, C10, C89, ChatGPT, D83, L86, Large language models, Political bias, Z00},
}

We investigate the political bias of a large language model (LLM), ChatGPT, which has become popular for retrieving factual information and generating content. Although ChatGPT assures that it is impartial, the literature suggests that LLMs exhibit bias involving race, gender, religion, and political orientation. Political bias in LLMs can have adverse political and electoral consequences similar to bias from traditional and social media. Moreover, political bias can be harder to detect and eradicate than gender or racial bias. We propose a novel empirical design to infer whether ChatGPT has political biases by requesting it to impersonate someone from a given side of the political spectrum and comparing these answers with its default. We also propose dose-response, placebo, and profession-politics alignment robustness tests. To reduce concerns about the randomness of the generated text, we collect answers to the same questions 100 times, with question order randomized on each round. We find robust evidence that ChatGPT presents a significant and systematic political bias toward the Democrats in the US, Lula in Brazil, and the Labour Party in the UK. These results translate into real concerns that ChatGPT, and LLMs in general, can extend or even amplify the existing challenges involving political processes posed by the Internet and social media. Our findings have important implications for policymakers, media, politics, and academia stakeholders.

@misc{gallegos_bias_2023,
	title = {Bias and {Fairness} in {Large} {Language} {Models}: {A} {Survey}},
	shorttitle = {Bias and {Fairness} in {Large} {Language} {Models}},
	url = {http://arxiv.org/abs/2309.00770},
	doi = {10.48550/arXiv.2309.00770},
	abstract = {Rapid advancements of large language models (LLMs) have enabled the processing, understanding, and generation of human-like text, with increasing integration into systems that touch our social sphere. Despite this success, these models can learn, perpetuate, and amplify harmful social biases. In this paper, we present a comprehensive survey of bias evaluation and mitigation techniques for LLMs. We first consolidate, formalize, and expand notions of social bias and fairness in natural language processing, defining distinct facets of harm and introducing several desiderata to operationalize fairness for LLMs. We then unify the literature by proposing three intuitive taxonomies, two for bias evaluation, namely metrics and datasets, and one for mitigation. Our first taxonomy of metrics for bias evaluation disambiguates the relationship between metrics and evaluation datasets, and organizes metrics by the different levels at which they operate in a model: embeddings, probabilities, and generated text. Our second taxonomy of datasets for bias evaluation categorizes datasets by their structure as counterfactual inputs or prompts, and identifies the targeted harms and social groups; we also release a consolidation of publicly-available datasets for improved access. Our third taxonomy of techniques for bias mitigation classifies methods by their intervention during pre-processing, in-training, intra-processing, and post-processing, with granular subcategories that elucidate research trends. Finally, we identify open problems and challenges for future work. Synthesizing a wide range of recent research, we aim to provide a clear guide of the existing literature that empowers researchers and practitioners to better understand and prevent the propagation of bias in LLMs.},
	urldate = {2023-09-15},
	publisher = {arXiv},
	author = {Gallegos, Isabel O. and Rossi, Ryan A. and Barrow, Joe and Tanjim, Md Mehrab and Kim, Sungchul and Dernoncourt, Franck and Yu, Tong and Zhang, Ruiyi and Ahmed, Nesreen K.},
	month = sep,
	year = {2023},
	note = {arXiv:2309.00770 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Computers and Society, Computer Science - Machine Learning},
}

Rapid advancements of large language models (LLMs) have enabled the processing, understanding, and generation of human-like text, with increasing integration into systems that touch our social sphere. Despite this success, these models can learn, perpetuate, and amplify harmful social biases. In this paper, we present a comprehensive survey of bias evaluation and mitigation techniques for LLMs. We first consolidate, formalize, and expand notions of social bias and fairness in natural language processing, defining distinct facets of harm and introducing several desiderata to operationalize fairness for LLMs. We then unify the literature by proposing three intuitive taxonomies, two for bias evaluation, namely metrics and datasets, and one for mitigation. Our first taxonomy of metrics for bias evaluation disambiguates the relationship between metrics and evaluation datasets, and organizes metrics by the different levels at which they operate in a model: embeddings, probabilities, and generated text. Our second taxonomy of datasets for bias evaluation categorizes datasets by their structure as counterfactual inputs or prompts, and identifies the targeted harms and social groups; we also release a consolidation of publicly-available datasets for improved access. Our third taxonomy of techniques for bias mitigation classifies methods by their intervention during pre-processing, in-training, intra-processing, and post-processing, with granular subcategories that elucidate research trends. Finally, we identify open problems and challenges for future work. Synthesizing a wide range of recent research, we aim to provide a clear guide of the existing literature that empowers researchers and practitioners to better understand and prevent the propagation of bias in LLMs.

@incollection{bewersdorff_tum-digillab_2023,
	address = {Wiesbaden},
	series = {Edition {Fachdidaktiken}},
	title = {Das {TUM}-{DigiLLab}: {Lehr}-{Lernraum} sowie {Forschungs}- und {Entwicklungsumgebung} zur {Förderung} digitaler {Kompetenzen}},
	isbn = {978-3-658-40109-2},
	shorttitle = {Das {TUM}-{DigiLLab}},
	url = {https://doi.org/10.1007/978-3-658-40109-2_10},
	doi = {10.1007/978-3-658-40109-2_10},
	abstract = {Das Digitale Lehr-Lern-Labor der Technischen Universität München (TUM-DigiLLab) soll als Ort der Entwicklung, Durchführung und Beforschung von Lehr-Lernkonzepten in authentischen Anwendungskontexten helfen die Lücke zwischen theoretischer Wissensvermittlung an der Universität und der Wissensanwendung in lebensweltlichen oder professionellen Zusammenhängen der Praxis zu schließen. Einen didaktischen Schwerpunkt bildet die Entwicklung innovativer Lehr-Lernkonzepte im Themenfeld der Künstlichen Intelligenz.},
	language = {de},
	urldate = {2023-09-15},
	booktitle = {Lehr-{Lern}-{Labore} und {Digitalisierung}},
	publisher = {Springer Fachmedien},
	author = {Bewersdorff, Arne and Nerdel, Claudia},
	editor = {Meier, Monique and Greefrath, Gilbert and Hammann, Marcus and Wodzinski, Rita and Ziepprecht, Kathrin},
	year = {2023},
	keywords = {Augmented Reality, Digitalisierung, Künstliche Intelligenz, Lehr-Lern-Labore, Technologiegestützte Kooperation},
	pages = {137--141},
}

Das Digitale Lehr-Lern-Labor der Technischen Universität München (TUM-DigiLLab) soll als Ort der Entwicklung, Durchführung und Beforschung von Lehr-Lernkonzepten in authentischen Anwendungskontexten helfen die Lücke zwischen theoretischer Wissensvermittlung an der Universität und der Wissensanwendung in lebensweltlichen oder professionellen Zusammenhängen der Praxis zu schließen. Einen didaktischen Schwerpunkt bildet die Entwicklung innovativer Lehr-Lernkonzepte im Themenfeld der Künstlichen Intelligenz.

@article{bewersdorff_myths_2023,
	title = {Myths, mis- and preconceptions of artificial intelligence: {A} review of the literature},
	volume = {4},
	issn = {2666-920X},
	shorttitle = {Myths, mis- and preconceptions of artificial intelligence},
	url = {https://www.sciencedirect.com/science/article/pii/S2666920X2300022X},
	doi = {10.1016/j.caeai.2023.100143},
	abstract = {Artificial Intelligence (AI) is prevalent in nearly every aspect of our lives. However, recent studies have found a significant amount of confusion and misunderstanding surrounding AI. To develop effective educational programs in the field of AI, it is vital to examine and understand learners' pre- and misconceptions as well as myths about AI. This study examined a corpus of 591 studies. 25 relevant studies were identified by applying the following eligibility criteria: English-written original empirical research on education and AI and reporting AI conceptions in a formal learning context. The review found studies from six continents, with the majority conducted in Europe and North America. The studies predominantly focus on the school and university levels. Findings reveal a range of preconceptions, misconceptions, and myths about AI, such as: Learners often have limited understanding of AI on a technical level. They tend to attribute human-like characteristics or attributes to AI systems and may have narrow views of AI's scope, capabilities, and limitations. The review also shows that learners often have binary and unspecific views about the threats, dangers, and benefits of AI. Effective educational programs are key to empower learners' understanding of AI, thus helping them make informed decisions about the integration of AI in our society, rather than being swayed by misinformation and unnecessary fear. This review may help inform the development of more effective teaching and outreach strategies in AI education.},
	urldate = {2023-09-15},
	journal = {Computers and Education: Artificial Intelligence},
	author = {Bewersdorff, Arne and Zhai, Xiaoming and Roberts, Jessica and Nerdel, Claudia},
	month = jan,
	year = {2023},
	keywords = {Artificial intelligence, Misconceptions, Preconceptions, Review},
	pages = {100143},
}

Artificial Intelligence (AI) is prevalent in nearly every aspect of our lives. However, recent studies have found a significant amount of confusion and misunderstanding surrounding AI. To develop effective educational programs in the field of AI, it is vital to examine and understand learners' pre- and misconceptions as well as myths about AI. This study examined a corpus of 591 studies. 25 relevant studies were identified by applying the following eligibility criteria: English-written original empirical research on education and AI and reporting AI conceptions in a formal learning context. The review found studies from six continents, with the majority conducted in Europe and North America. The studies predominantly focus on the school and university levels. Findings reveal a range of preconceptions, misconceptions, and myths about AI, such as: Learners often have limited understanding of AI on a technical level. They tend to attribute human-like characteristics or attributes to AI systems and may have narrow views of AI's scope, capabilities, and limitations. The review also shows that learners often have binary and unspecific views about the threats, dangers, and benefits of AI. Effective educational programs are key to empower learners' understanding of AI, thus helping them make informed decisions about the integration of AI in our society, rather than being swayed by misinformation and unnecessary fear. This review may help inform the development of more effective teaching and outreach strategies in AI education.

@book{hose_formen_2023,
	address = {Stuttgart},
	title = {Formen und {Funktionen} griechisch-römischer {Literatur} : {Aufsätze} zur {Literaturgeschichte} und {Literaturgeschichtsschreibung}},
	isbn = {978-3-515-13411-8},
	shorttitle = {Formen und {Funktionen} griechisch-römischer {Literatur}},
	url = {https://d-nb.info/1273972600/04},
	publisher = {Franz Steiner Verlag},
	author = {Hose, Martin},
	editor = {Peri, Annamaria and Thum, Tobias},
	year = {2023},
}

@book{baker_chatgpt_2023,
	title = {{ChatGPT} für {Dummies}},
	isbn = {978-3-527-84473-9},
	abstract = {Profitieren auch Sie von den wunderbaren Fähigkeiten von ChatGPT. Pam Baker erklärt Ihnen, wie ChatGPT funktioniert und wie Sie den Chatbot gewinnbringend einsetzen - sei es bei der Texterstellung für Werbezwecke, der Kundenbetreuung auf einer Webseite oder für die Beantwortung all jener Fragen, auf die Sie bisher keine Antwort gefunden haben. Sie lernen die Stärken und Schwächen des Tools kennen. So können Sie besser einschätzen, wo es Ihnen nutzt und wo Sie besser weiter arbeiten wie bisher. Erschließen Sie das Potenzial von ChatGPT!},
	language = {de},
	publisher = {John Wiley \& Sons},
	author = {Baker, Pam},
	month = aug,
	year = {2023},
	note = {Google-Books-ID: dEbREAAAQBAJ},
	keywords = {Computers / Artificial Intelligence / General, Computers / Computer Science, Computers / Information Technology},
}

Profitieren auch Sie von den wunderbaren Fähigkeiten von ChatGPT. Pam Baker erklärt Ihnen, wie ChatGPT funktioniert und wie Sie den Chatbot gewinnbringend einsetzen - sei es bei der Texterstellung für Werbezwecke, der Kundenbetreuung auf einer Webseite oder für die Beantwortung all jener Fragen, auf die Sie bisher keine Antwort gefunden haben. Sie lernen die Stärken und Schwächen des Tools kennen. So können Sie besser einschätzen, wo es Ihnen nutzt und wo Sie besser weiter arbeiten wie bisher. Erschließen Sie das Potenzial von ChatGPT!

@misc{zhao_survey_2023,
	title = {A {Survey} of {Large} {Language} {Models}},
	url = {http://arxiv.org/abs/2303.18223},
	abstract = {Ever since the Turing Test was proposed in the 1950s, humans have explored the mastering of language intelligence by machine. Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable artificial intelligence (AI) algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pretraining Transformer models over large-scale corpora, showing strong capabilities in solving various natural language processing (NLP) tasks. Since the researchers have found that model scaling can lead to an improved model capacity, they further investigate the scaling effect by increasing the parameter scale to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement, but also exhibit some special abilities (e.g., incontext learning) that are not present in small-scale language models (e.g., BERT). To discriminate the language models in different parameter scales, the research community has coined the term large language models (LLM) for the PLMs of significant size (e.g., containing tens or hundreds of billions of parameters). Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT (a powerful AI chatbot developed based on LLMs), which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. Considering this rapid technical progress, in this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Furthermore, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions. This survey provides an up-to-date review of the literature on LLMs, which can be a useful resource for both researchers and engineers.},
	language = {en},
	urldate = {2023-09-14},
	publisher = {arXiv},
	author = {Zhao, Wayne Xin and Zhou, Kun and Li, Junyi and Tang, Tianyi and Wang, Xiaolei and Hou, Yupeng and Min, Yingqian and Zhang, Beichen and Zhang, Junjie and Dong, Zican and Du, Yifan and Yang, Chen and Chen, Yushuo and Chen, Zhipeng and Jiang, Jinhao and Ren, Ruiyang and Li, Yifan and Tang, Xinyu and Liu, Zikang and Liu, Peiyu and Nie, Jian-Yun and Wen, Ji-Rong},
	month = sep,
	year = {2023},
	note = {arXiv:2303.18223 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language},
}

Ever since the Turing Test was proposed in the 1950s, humans have explored the mastering of language intelligence by machine. Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable artificial intelligence (AI) algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pretraining Transformer models over large-scale corpora, showing strong capabilities in solving various natural language processing (NLP) tasks. Since the researchers have found that model scaling can lead to an improved model capacity, they further investigate the scaling effect by increasing the parameter scale to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement, but also exhibit some special abilities (e.g., incontext learning) that are not present in small-scale language models (e.g., BERT). To discriminate the language models in different parameter scales, the research community has coined the term large language models (LLM) for the PLMs of significant size (e.g., containing tens or hundreds of billions of parameters). Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT (a powerful AI chatbot developed based on LLMs), which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. Considering this rapid technical progress, in this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Furthermore, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions. This survey provides an up-to-date review of the literature on LLMs, which can be a useful resource for both researchers and engineers.

@article{linka_pain_2023,
	title = {Pain in {Classical} {Greek} {Texts}},
	copyright = {Copyright (c) 2023},
	issn = {2364-7957},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/93792},
	doi = {10.11588/dco.2023.9.93792},
	abstract = {Texte aus der klassischen griechischen Periode spielen eine entscheidende Rolle in der historischen Entwicklung der westlichen Wissenschaft und Philosophie. Das Konzept des Schmerzes ist vor allem für zwei Bereiche des menschlichen Wissens, nämlich Medizin und Ethik, von zentraler Bedeutung. Obwohl der Begriff des Schmerzes für beide Bereiche wichtig ist, werden sie in der Wissenschaft meist getrennt voneinander untersucht. Wir betrachten sie gemeinsam, im Kontext der gesamten erhaltenen klassischen griechischen Literatur. Dies wird durch unseren methodischen Ansatz ermöglicht, der traditionelle Interpretationsansätze mit computergestützten Textanalysemethoden kombiniert und so die Untersuchung einer großen Menge von Textdaten ermöglicht. Wenn wir den Kontext der Verwendung einzelner Wörter, die Schmerz bezeichnen, in den Texten verschiedener Gattungen oder Themen betrachten, können wir relativ stabile semantische Cluster identifizieren, auf die auf Schmerz hindeuten, wie etwa Pathologien, Emotionen oder Moral. Auf diese Weise sind wir in der Lage, die Rolle bestimmter Schmerzwörter, ihre Bedeutung und ihre wechselseitigen Beziehungen in klassischen griechischen Texten zu erfassen. Unser Ansatz ermöglicht es uns auch, die Rolle verschiedener textueller Subkorpora (philosophisch, medizinisch) für die Art und Weise zu erkennen, wie Schmerz in klassischen griechischen Texten aufgefasst wurde.},
	language = {en},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Linka, Vojtěch and Kaše, Vojtěch},
	month = apr,
	year = {2023},
	keywords = {Semantische Analyse},
	pages = {1--14},
}

Texte aus der klassischen griechischen Periode spielen eine entscheidende Rolle in der historischen Entwicklung der westlichen Wissenschaft und Philosophie. Das Konzept des Schmerzes ist vor allem für zwei Bereiche des menschlichen Wissens, nämlich Medizin und Ethik, von zentraler Bedeutung. Obwohl der Begriff des Schmerzes für beide Bereiche wichtig ist, werden sie in der Wissenschaft meist getrennt voneinander untersucht. Wir betrachten sie gemeinsam, im Kontext der gesamten erhaltenen klassischen griechischen Literatur. Dies wird durch unseren methodischen Ansatz ermöglicht, der traditionelle Interpretationsansätze mit computergestützten Textanalysemethoden kombiniert und so die Untersuchung einer großen Menge von Textdaten ermöglicht. Wenn wir den Kontext der Verwendung einzelner Wörter, die Schmerz bezeichnen, in den Texten verschiedener Gattungen oder Themen betrachten, können wir relativ stabile semantische Cluster identifizieren, auf die auf Schmerz hindeuten, wie etwa Pathologien, Emotionen oder Moral. Auf diese Weise sind wir in der Lage, die Rolle bestimmter Schmerzwörter, ihre Bedeutung und ihre wechselseitigen Beziehungen in klassischen griechischen Texten zu erfassen. Unser Ansatz ermöglicht es uns auch, die Rolle verschiedener textueller Subkorpora (philosophisch, medizinisch) für die Art und Weise zu erkennen, wie Schmerz in klassischen griechischen Texten aufgefasst wurde.

@misc{burns_latincy_2023,
	title = {{LatinCy}: {Synthetic} {Trained} {Pipelines} for {Latin} {NLP}},
	shorttitle = {{LatinCy}},
	url = {http://arxiv.org/abs/2305.04365},
	doi = {10.48550/arXiv.2305.04365},
	abstract = {This paper introduces LatinCy, a set of trained general purpose Latin-language "core" pipelines for use with the spaCy natural language processing framework. The models are trained on a large amount of available Latin data, including all five of the Latin Universal Dependency treebanks, which have been preprocessed to be compatible with each other. The result is a set of general models for Latin with good performance on a number of natural language processing tasks (e.g. the top-performing model yields POS tagging, 97.41\% accuracy; lemmatization, 94.66\% accuracy; morphological tagging 92.76\% accuracy). The paper describes the model training, including its training data and parameterization, and presents the advantages to Latin-language researchers of having a spaCy model available for NLP work.},
	urldate = {2023-07-16},
	publisher = {arXiv},
	author = {Burns, Patrick J.},
	month = may,
	year = {2023},
	note = {arXiv:2305.04365 [cs]
version: 1},
	keywords = {Computer Science - Computation and Language},
}

This paper introduces LatinCy, a set of trained general purpose Latin-language "core" pipelines for use with the spaCy natural language processing framework. The models are trained on a large amount of available Latin data, including all five of the Latin Universal Dependency treebanks, which have been preprocessed to be compatible with each other. The result is a set of general models for Latin with good performance on a number of natural language processing tasks (e.g. the top-performing model yields POS tagging, 97.41% accuracy; lemmatization, 94.66% accuracy; morphological tagging 92.76% accuracy). The paper describes the model training, including its training data and parameterization, and presents the advantages to Latin-language researchers of having a spaCy model available for NLP work.

@article{clavert_publishing_2022,
	title = {Publishing digital history scholarship in the era of updatism},
	volume = {2},
	url = {https://doi.org/10.1515/JDH-2022-0003?locatt=label:JDHFULL},
	number = {1},
	journal = {Journal of Digital History},
	author = {Clavert, Frédéric and Fickers, Andreas},
	year = {2022},
}

@book{guiliano_primer_2022,
	title = {A {Primer} for {Teaching} {Digital} {History}: {Ten} {Design} {Principles}},
	isbn = {978-1-4780-1505-5},
	url = {http://www.jstor.org/stable/j.ctv2j86bhk},
	abstract = {A Primer for Teaching Digital History is a guide for college and high school teachers who are teaching digital history for the first time or for experienced teachers who want to reinvigorate their pedagogy. It can also serve those who are training future teachers to prepare their own syllabi, as well as teachers who want to incorporate digital history into their history courses. Offering design principles for approaching digital history that represent the possibilities that digital research and scholarship can take, Jennifer Guiliano outlines potential strategies and methods for building syllabi and curricula. Taking readers through the process of selecting data, identifying learning outcomes, and determining which tools students will use in the classroom, Guiliano outlines popular research methods including digital source criticism, text analysis, and visualization. She also discusses digital archives, exhibits, and collections as well as audiovisual and mixed-media narratives such as short documentaries, podcasts, and multimodal storytelling. Throughout, Guiliano illuminates how digital history can enhance understandings of not just what histories are told but how they are told and who has access to them.},
	urldate = {2026-01-25},
	publisher = {Duke University Press},
	author = {Guiliano, Jennifer},
	year = {2022},
}

A Primer for Teaching Digital History is a guide for college and high school teachers who are teaching digital history for the first time or for experienced teachers who want to reinvigorate their pedagogy. It can also serve those who are training future teachers to prepare their own syllabi, as well as teachers who want to incorporate digital history into their history courses. Offering design principles for approaching digital history that represent the possibilities that digital research and scholarship can take, Jennifer Guiliano outlines potential strategies and methods for building syllabi and curricula. Taking readers through the process of selecting data, identifying learning outcomes, and determining which tools students will use in the classroom, Guiliano outlines popular research methods including digital source criticism, text analysis, and visualization. She also discusses digital archives, exhibits, and collections as well as audiovisual and mixed-media narratives such as short documentaries, podcasts, and multimodal storytelling. Throughout, Guiliano illuminates how digital history can enhance understandings of not just what histories are told but how they are told and who has access to them.

@incollection{corfield_using_2022,
	title = {Using technology creatively: digital history},
	isbn = {978-1-914477-15-7},
	url = {http://www.jstor.org/stable/j.ctv2kcwp3s.16},
	abstract = {Whether the research project focuses upon the history of poetry or economics – whether it is about gender, discourse, diplomacy or cash flows – the way historians analyse their sources has changed substantially in the last thirty years.¹ As a large (though selective) proportion of inherited print and manuscripts have been digitized, all historians have become ‘digital’. They access many of their sources as digital files and write history while sitting in front of a screen.Furthermore, ‘born digital’ sources are increasingly important, and images that were traditionally inaccessible in museum collections are now much more available – and available for analysis in},
	urldate = {2026-01-25},
	booktitle = {Becoming a {Historian}: {An} {Informal} {Guide}},
	publisher = {University of London Press},
	author = {Corfield, Penelope J. and Hitchcock, Tim},
	year = {2022},
	pages = {93--102},
}

Whether the research project focuses upon the history of poetry or economics – whether it is about gender, discourse, diplomacy or cash flows – the way historians analyse their sources has changed substantially in the last thirty years.¹ As a large (though selective) proportion of inherited print and manuscripts have been digitized, all historians have become ‘digital’. They access many of their sources as digital files and write history while sitting in front of a screen.Furthermore, ‘born digital’ sources are increasingly important, and images that were traditionally inaccessible in museum collections are now much more available – and available for analysis in

@article{verplanken_attitudes_2022,
	title = {Attitudes, {Habits}, and {Behavior} {Change}},
	volume = {73},
	issn = {1545-2085},
	doi = {10.1146/annurev-psych-020821-011744},
	abstract = {Efforts to guide peoples' behavior toward environmental sustainability, good health, or new products have emphasized informational and attitude change strategies. There is evidence that changing attitudes leads to changes in behavior, yet this approach takes insufficient account of the nature and operation of habits, which form boundary conditions for attitude-directed interventions. Integration of research on attitudes and habits might enable investigators to identify when and how behavior change strategies will be most effective. How might attitudinally driven behavior change be consolidated into lasting habits? How do habits protect the individual against the vicissitudes of attitudes and temptations and promote goal achievement? How might attitudinal approaches aiming to change habits be improved by capitalizing on habit discontinuities and strategic planning? When and how might changing or creating habit architecture shape habits directly? A systematic approach to these questions might help move behavior change efforts from attitude change strategies to habit change strategies.},
	language = {eng},
	journal = {Annual Review of Psychology},
	author = {Verplanken, Bas and Orbell, Sheina},
	month = jan,
	year = {2022},
	keywords = {Attitude, Habits, Health Behavior, Humans, Motivation, attitude–behavior relation, automatic processes, habit architecture, habit discontinuity, habit formation, self-regulation},
	pages = {327--352},
}

Efforts to guide peoples' behavior toward environmental sustainability, good health, or new products have emphasized informational and attitude change strategies. There is evidence that changing attitudes leads to changes in behavior, yet this approach takes insufficient account of the nature and operation of habits, which form boundary conditions for attitude-directed interventions. Integration of research on attitudes and habits might enable investigators to identify when and how behavior change strategies will be most effective. How might attitudinally driven behavior change be consolidated into lasting habits? How do habits protect the individual against the vicissitudes of attitudes and temptations and promote goal achievement? How might attitudinal approaches aiming to change habits be improved by capitalizing on habit discontinuities and strategic planning? When and how might changing or creating habit architecture shape habits directly? A systematic approach to these questions might help move behavior change efforts from attitude change strategies to habit change strategies.

@book{vuorikari_digcomp_2022,
	address = {JRC Publications Repository},
	title = {{DigComp} 2.2: {The} {Digital} {Competence} {Framework} for {Citizens} - {With} new examples of knowledge, skills and attitudes},
	isbn = {978-92-76-48882-8},
	issn = {1831-9424, 1018-5593},
	shorttitle = {{DigComp} 2.2},
	url = {https://publications.jrc.ec.europa.eu/repository/handle/JRC128415},
	doi = {10.2760/115376},
	abstract = {The Digital Competence Framework for Citizen (DigComp) provides a common understanding of what digital competence is. The present publication has two main parts: the integrated DigComp 2.2 framework provides more than 250 new examples of knowledge, skills and attitudes that help citizens engage confidently, critically and safely with digital technologies, and new and emerging ones such as systems driven by artificial intelligence (AI). The framework is also made available following the digital accessibility guidelines, as creating accessible digital resources is an important priority today. The second part of the publication gives a snapshot of the existing reference material for DigComp consolidating previously released publications and references.},
	language = {en},
	urldate = {2025-04-01},
	author = {Vuorikari, Riina and Kluzer, Stefano and Punie, Yves},
	year = {2022},
}

The Digital Competence Framework for Citizen (DigComp) provides a common understanding of what digital competence is. The present publication has two main parts: the integrated DigComp 2.2 framework provides more than 250 new examples of knowledge, skills and attitudes that help citizens engage confidently, critically and safely with digital technologies, and new and emerging ones such as systems driven by artificial intelligence (AI). The framework is also made available following the digital accessibility guidelines, as creating accessible digital resources is an important priority today. The second part of the publication gives a snapshot of the existing reference material for DigComp consolidating previously released publications and references.

@article{assael_restoring_2022,
	title = {Restoring and attributing ancient texts using deep neural networks},
	volume = {603},
	copyright = {2022 The Author(s)},
	issn = {1476-4687},
	url = {https://www.nature.com/articles/s41586-022-04448-z},
	doi = {10.1038/s41586-022-04448-z},
	abstract = {Ancient history relies on disciplines such as epigraphy—the study of inscribed texts known as inscriptions—for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian’s workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62\% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25\% to 72\%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71\% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.},
	language = {en},
	number = {7900},
	urldate = {2022-03-21},
	journal = {Nature},
	publisher = {Nature Publishing Group},
	author = {Assael, Yannis and Sommerschield, Thea and Shillingford, Brendan and Bordbar, Mahyar and Pavlopoulos, John and Chatzipanagiotou, Marita and Androutsopoulos, Ion and Prag, Jonathan and de Freitas, Nando},
	month = mar,
	year = {2022},
	note = {Number: 7900},
	keywords = {Archaeology, Computer science, History},
	pages = {280--283},
}

Ancient history relies on disciplines such as epigraphy—the study of inscribed texts known as inscriptions—for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian’s workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25% to 72%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.

@article{dorpinghaus_digital_2022,
	title = {Digital theology: new perspectives on interdisciplinary research between the humanities and theology},
	volume = {18},
	issn = {1556-3723},
	url = {https://www.religjournal.com/pdf/ijrr18002.pdf},
	journal = {Interdisciplinary Journal of Research on Religion},
	publisher = {Baylor University},
	author = {Dörpinghaus, Jens},
	year = {2022},
}

@article{voda_exploring_2022,
	title = {Exploring {Digital} {Literacy} {Skills} in {Social} {Sciences} and {Humanities} {Students}},
	volume = {14},
	copyright = {http://creativecommons.org/licenses/by/3.0/},
	issn = {2071-1050},
	url = {https://www.mdpi.com/2071-1050/14/5/2483},
	doi = {10.3390/su14052483},
	abstract = {Digital literacy is among the mandatory abilities to any higher education level and represents a fundamental ingredient in successful professionalization. Considering the deep penetration of digital technologies in everyday life, digital literacy offers a set of transversal skills that could improve a whole area of activities, from banking operations to civic participation. However, these skills are diverse and vary according to the development of technologies and society. This study fills an important academic gap on digital literacy by placing it in a specific and well-defined context, analyzing different perspectives that involve such learning, such as predictors of digital literacy in different types of students. In addition, research increases its importance as it is being developed during the pandemic, a period characterized by accelerated technological use and sudden changes. This research used a quantitative design based on the answers to a questionnaire conducted from March 2021 to May 2021. From a methodological perspective, we tested several hypotheses using one-way analysis of variance (ANOVA) and confirmatory factor analysis (CFA) within the structural equation model (SEM). The results show that communication, critical thinking, problem-solving, and technical digital skills are more present in the case of students enrolled in social sciences, while other digital skills (i.e., creativity and information) are more prevalent in the case of humanities students. Moreover, the results showed that, except for creativity and problem-solving-related digital skills, all of the digital skills were significantly influenced by students’ different levels of education.},
	language = {en},
	number = {5},
	urldate = {2025-10-23},
	journal = {Sustainability},
	publisher = {Multidisciplinary Digital Publishing Institute},
	author = {Vodă, Ana Iolanda and Cautisanu, Cristina and Grădinaru, Camelia and Tănăsescu, Chris and de Moraes, Gustavo Herminio Salati Marcondes},
	month = jan,
	year = {2022},
	keywords = {COVID-19, digital course enrolment, digital literacy, digital skills, education level},
	pages = {2483},
}

Digital literacy is among the mandatory abilities to any higher education level and represents a fundamental ingredient in successful professionalization. Considering the deep penetration of digital technologies in everyday life, digital literacy offers a set of transversal skills that could improve a whole area of activities, from banking operations to civic participation. However, these skills are diverse and vary according to the development of technologies and society. This study fills an important academic gap on digital literacy by placing it in a specific and well-defined context, analyzing different perspectives that involve such learning, such as predictors of digital literacy in different types of students. In addition, research increases its importance as it is being developed during the pandemic, a period characterized by accelerated technological use and sudden changes. This research used a quantitative design based on the answers to a questionnaire conducted from March 2021 to May 2021. From a methodological perspective, we tested several hypotheses using one-way analysis of variance (ANOVA) and confirmatory factor analysis (CFA) within the structural equation model (SEM). The results show that communication, critical thinking, problem-solving, and technical digital skills are more present in the case of students enrolled in social sciences, while other digital skills (i.e., creativity and information) are more prevalent in the case of humanities students. Moreover, the results showed that, except for creativity and problem-solving-related digital skills, all of the digital skills were significantly influenced by students’ different levels of education.

@article{liebsch_fur_2022,
	title = {Für eine {Partizipationskultur} – {Inklusion} und {Digitalisierung} in den {Alten} {Sprachen}},
	volume = {3},
	issn = {1432-7511},
	doi = {https://doi.org/10.11588/fc.2022.3.92533},
	number = {65},
	journal = {Forum Classicum},
	author = {Liebsch, Ann-Catherine},
	year = {2022},
	pages = {238--245},
}

@article{laupichler_artificial_2022,
	title = {Artificial intelligence literacy in higher and adult education: {A} scoping literature review},
	volume = {3},
	issn = {2666-920X},
	shorttitle = {Artificial intelligence literacy in higher and adult education},
	url = {https://www.sciencedirect.com/science/article/pii/S2666920X2200056X},
	doi = {10.1016/j.caeai.2022.100101},
	abstract = {Since artificial intelligence (AI) is finding its way into more and more areas of everyday life, improving the AI skills of non-experts is important and will become even more relevant in the future. While it is necessary that children learn about the possibilities of AI at an early age, adults in higher education and beyond should also have at least a basic understanding of AI (i.e., AI literacy) to be able to interact effectively with the technology. To evaluate the current state of the literature on AI literacy in higher and adult education, a scoping literature review was conducted with the objective of identifying thematic foci and recent research trends. Ten research databases were searched and out of 902 initial records, 30 studies were identified using predefined eligibility criteria, whereof the content was evaluated in the review. The results indicated that research in this area is still in its infancy and needs refinement in terms of how to define AI literacy in adult education as well as what content should be taught to non-experts. Lastly, several recommendations for research and practice are derived from the results.},
	urldate = {2023-12-21},
	journal = {Computers and Education: Artificial Intelligence},
	author = {Laupichler, Matthias Carl and Aster, Alexandra and Schirch, Jana and Raupach, Tobias},
	month = jan,
	year = {2022},
	keywords = {AI education, AI literacy, Higher education, Literature review, Teaching AI},
	pages = {100101},
}

Since artificial intelligence (AI) is finding its way into more and more areas of everyday life, improving the AI skills of non-experts is important and will become even more relevant in the future. While it is necessary that children learn about the possibilities of AI at an early age, adults in higher education and beyond should also have at least a basic understanding of AI (i.e., AI literacy) to be able to interact effectively with the technology. To evaluate the current state of the literature on AI literacy in higher and adult education, a scoping literature review was conducted with the objective of identifying thematic foci and recent research trends. Ten research databases were searched and out of 902 initial records, 30 studies were identified using predefined eligibility criteria, whereof the content was evaluated in the review. The results indicated that research in this area is still in its infancy and needs refinement in terms of how to define AI literacy in adult education as well as what content should be taught to non-experts. Lastly, several recommendations for research and practice are derived from the results.

@article{schroter_validating_2022,
	title = {Validating {Topic} {Modeling} as a {Method} of {Analyzing} {Sujet} and {Theme}},
	volume = {1},
	copyright = {Copyright: © 2022 The Author(s). The text of this work is released under the Creative Commons license CC BY 4.0 International. You can find the contract text of the license at https://creativecommons.org/licenses/by/4.0/. The illustrations are excluded from this license, here the copyright lies with the respective rights holder.},
	issn = {2940-1348},
	url = {https://jcls.io/article/id/91/},
	doi = {10.48694/jcls.91},
	abstract = {In Computational Literary Studies (CLS), several procedures for thematic analysis have been adapted from NLP and Computer Science. Among these procedures, topic modeling is the most prominent and popular technique. We maintain, however, that this procedure is used only in the context of exploration up to date, but not in the context of justification. When we seek to prove assumptions concerning the correlation between genres, methods of computational text analysis have to be set up in research environments of justification, i.e. in environments of hypothesis testing. We provide a holistic model of validation and conceptual disambiguation of the notion of aboutness as sujet, fabula, and theme, and discuss essential methodological requirements for hypothesis-based analysis. As we maintain that validation has to be performed for individual tasks respectively, we shall perform empirical validation of topic modeling based on a new corpus of German novellas and comprehensive annotations and draw hypothetical generalizations on the applicability of topic modeling for analyzing aboutness in the domain of narrative fiction.},
	language = {eng},
	number = {1},
	urldate = {2025-07-07},
	journal = {Journal of Computational Literary Studies},
	publisher = {Universitäts- und Landesbibliothek Darmstadt},
	author = {Schröter, Julian and Du, Keli},
	month = dec,
	year = {2022},
	note = {Number: 1},
}

In Computational Literary Studies (CLS), several procedures for thematic analysis have been adapted from NLP and Computer Science. Among these procedures, topic modeling is the most prominent and popular technique. We maintain, however, that this procedure is used only in the context of exploration up to date, but not in the context of justification. When we seek to prove assumptions concerning the correlation between genres, methods of computational text analysis have to be set up in research environments of justification, i.e. in environments of hypothesis testing. We provide a holistic model of validation and conceptual disambiguation of the notion of aboutness as sujet, fabula, and theme, and discuss essential methodological requirements for hypothesis-based analysis. As we maintain that validation has to be performed for individual tasks respectively, we shall perform empirical validation of topic modeling based on a new corpus of German novellas and comprehensive annotations and draw hypothetical generalizations on the applicability of topic modeling for analyzing aboutness in the domain of narrative fiction.

@inproceedings{sprugnoli_sentiment_2022,
	title = {Sentiment {Analysis} of {Latin} {Poetry}: {First} {Experiments} on the {Odes} of {Horace}},
	url = {https://books.openedition.org/aaccademia/10854},
	booktitle = {Proceedings of the {Eighth} {Italian} {Conference} on {Computational} {Linguistics} {CliC}-{It} 2021},
	author = {Sprugnoli, Rachele and Mambrini, Francesco and Passarotti, Marco and Moretti, Giovanni},
	year = {2022},
}

@inproceedings{yamshchikov_bert_2022,
	title = {{BERT} in {Plutarch}'s {Shadows}},
	url = {https://aclanthology.org/2022.emnlp-main.407.pdf},
	doi = {10.18653/v1/2022.emnlp-main.407},
	booktitle = {Proceedings of the 2022 {Conference} on {Empirical} {Methods} in {Natural} {Language} {Processing}},
	publisher = {Association for Computational Linguistics},
	author = {Yamshchikov, Ivan P and Tikhonov, Alexey and Pantis, Yorgos and Schubert, Charlotte and Jost, Jürgen},
	year = {2022},
	pages = {6071--6080},
}

@article{assael_restoring_2022,
	title = {Restoring and attributing ancient texts using deep neural networks},
	volume = {603},
	copyright = {2022 The Author(s)},
	issn = {1476-4687},
	url = {https://www.nature.com/articles/s41586-022-04448-z},
	doi = {10.1038/s41586-022-04448-z},
	abstract = {Ancient history relies on disciplines such as epigraphy—the study of inscribed texts known as inscriptions—for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian’s workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62\% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25\% to 72\%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71\% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.},
	language = {en},
	number = {7900},
	urldate = {2023-04-26},
	journal = {Nature},
	publisher = {Nature Publishing Group},
	author = {Assael, Yannis and Sommerschield, Thea and Shillingford, Brendan and Bordbar, Mahyar and Pavlopoulos, John and Chatzipanagiotou, Marita and Androutsopoulos, Ion and Prag, Jonathan and de Freitas, Nando},
	month = mar,
	year = {2022},
	note = {Number: 7900},
	keywords = {Archaeology, Computer science, History},
	pages = {280--283},
}

Ancient history relies on disciplines such as epigraphy—the study of inscribed texts known as inscriptions—for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian’s workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25% to 72%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.

@article{mcgillivray_new_2022,
	title = {A {New} {Corpus} {Annotation} {Framework} for {Latin} {Diachronic} {Lexical} {Semantics}},
	volume = {21},
	url = {doi:10.1515/joll-2022-2007.},
	number = {1},
	journal = {Journal of Latin linguistics},
	author = {McGillivray, Barbara and Kondakova, Daria and Burman, Annie and Dell’Oro, Francesca and Bermúdez Sabel, Helena and Marongiu, Paola and Cruz, Manuel Márquez},
	year = {2022},
	keywords = {Latin lexical semantics, LatinISE corpus, annotation, semantic change},
	pages = {47--105},
}

@inproceedings{pavlopoulos_sentiment_2022,
	title = {Sentiment {Analysis} of {Homeric} {Text}: {The} 1st {Book} of {Iliad}},
	url = {https://aclanthology.org/2022.lrec-1.765.pdf},
	author = {Pavlopoulos, John and Xenos, Alexandros and Picca, Davide},
	year = {2022},
	pages = {7071--7077},
}

@incollection{de_graaf_agile_2022,
	address = {Marseille, France},
	title = {{AGILe}: {The} {First} {Lemmatizer} for {Ancient} {Greek} {Inscriptions}},
	booktitle = {Proceedings of the 13th {Conference} on {Language} {Resources} and {Evaluation} ({LREC} 2022)},
	author = {de Graaf, E. and Stopponi, S. and Bos, J. and Peels-Matthey, S. and Nissm, M.},
	year = {2022},
	pages = {5334--5344},
}

@article{prieto_espinosa_corpus_2022,
	title = {El {Corpus} {Documentale} {Latinum} {Hispaniarum} ({CODOLHisp}), una plataforma digital d’accés conjunt per a l’estudi del llatí medieval hispànic},
	url = {https://raco.cat/index.php/LlenguaLiteratura/article/view/399457},
	journal = {Llengua i literatura : revista anual de la Societat Catalana de Llengua i Literatura},
	author = {Prieto Espinosa, Carlos},
	year = {2022},
	pages = {204--207},
}

@article{forstall_towards_2022,
	title = {Towards a {Linked} {Open} {Data} {Resource} for {Direct} {Speech} {Acts} in {Greek} and {Latin} {Epic}},
	volume = {37},
	doi = {doi:10.1093/llc/fqac006.},
	number = {4},
	journal = {Digital scholarship in the humanities},
	author = {Forstall, Christopher W. and Finkmann, Simone and Verhelst, Berenice},
	year = {2022},
	pages = {972--981},
}

@book{regnault_annotation_2022,
	address = {Paris, Francehttps://hal-lirmm.ccsd.cnrs.fr/AO-LINGUISTIQUE/tel-04069848v1.},
	title = {Annotation et analyse syntaxique de corpus hétérogènes : le cas du français médiéval},
	url = {https://hal-lirmm.ccsd.cnrs.fr/AO-LINGUISTIQUE/tel-04069848v1},
	publisher = {Univ. de la Sorbonne Nouvelle (Paris III)},
	author = {Regnault, Mathilde},
	year = {2022},
}

@article{kenty_irony_2022,
	title = {Irony and {Figured} {Language} in {Cicero}’s {Letter} to {Lucceius}},
	volume = {118},
	language = {English},
	number = {1},
	journal = {Classical Journal},
	author = {Kenty, Joanna},
	year = {2022},
	pages = {50--89},
}

@inproceedings{sprugnoli_overview_2022,
	address = {Marseille, France},
	title = {Overview of the {EvaLatin} 2022 {Evaluation} {Campaign}},
	url = {https://aclanthology.org/2022.lt4hala-1.29},
	abstract = {This paper describes the organization and the results of the second edition of EvaLatin, the campaign for the evaluation of Natural Language Processing tools for Latin. The three shared tasks proposed in EvaLatin 2022, i.,e.,Lemmatization, Part-of-Speech Tagging and Features Identification, are aimed to foster research in the field of language technologies for Classical languages. The shared dataset consists of texts mainly taken from the LASLA corpus. More specifically, the training set includes only prose texts of the Classical period, whereas the test set is organized in three sub-tasks: a Classical sub-task on a prose text of an author not included in the training data, a Cross-genre sub-task on poetic and scientific texts, and a Cross-time sub-task on a text of the 15th century. The results obtained by the participants for each task and sub-task are presented and discussed.},
	urldate = {2023-10-06},
	booktitle = {Proceedings of the {Second} {Workshop} on {Language} {Technologies} for {Historical} and {Ancient} {Languages}},
	publisher = {European Language Resources Association},
	author = {Sprugnoli, Rachele and Passarotti, Marco and Cecchini, Flavio Massimiliano and Fantoli, Margherita and Moretti, Giovanni},
	month = jun,
	year = {2022},
	pages = {183--188},
}

This paper describes the organization and the results of the second edition of EvaLatin, the campaign for the evaluation of Natural Language Processing tools for Latin. The three shared tasks proposed in EvaLatin 2022, i.,e.,Lemmatization, Part-of-Speech Tagging and Features Identification, are aimed to foster research in the field of language technologies for Classical languages. The shared dataset consists of texts mainly taken from the LASLA corpus. More specifically, the training set includes only prose texts of the Classical period, whereas the test set is organized in three sub-tasks: a Classical sub-task on a prose text of an author not included in the training data, a Cross-genre sub-task on poetic and scientific texts, and a Cross-time sub-task on a text of the 15th century. The results obtained by the participants for each task and sub-task are presented and discussed.

@inproceedings{passarotti_issues_2022,
	title = {Issues in {Building} the {LiLa} {Knowledge} {Base} of {Interoperable} {Linguistic} {Resources} for {Latin}},
	url = {https://zenodo.org/doi/10.5281/zenodo.7263412},
	author = {Passarotti, Marco and Mambrini, Francesco},
	year = {2022},
}

@article{engelhardt_how_2022,
	title = {How to be {FAIR} with your data},
	copyright = {https://creativecommons.org/licenses/by/4.0/},
	url = {https://www.univerlag.uni-goettingen.de/handle/3/isbn-978-3-86395-539-7},
	doi = {10.17875/gup2022-1915},
	abstract = {Softcover, 17x24},
	language = {eng},
	urldate = {2023-10-02},
	author = {Engelhardt, Claudia and Barthauer, Raisa and Biernacka, Katarzyna and Coffey, Aoife and Cornet, Ronald and Danciu, Alina and Demchenko, Yuri and Downes, Stephen and Erdmann, Christopher and Garbuglia, Federica and Germer, Kerstin and Helbig, Kerstin and Hellström, Margareta and Hettne, Kristina and Hibbert, Dawn and Jetten, Mijke and Karimova, Yulia and Hansen, Karsten Kryger and Kuusniemi, Mari Elisa and Letizia, Viviana and McCutcheon, Valerie and McGillivray, Barbara and Ostrop, Jenny and Petersen, Britta and Petrus, Ana and Reichmann, Stefan and Rettberg, Najla and Reverté, Carmen and Rochlin, Nick and Saenen, Bregt and Schmidt, Birgit and Scholten, Jolien and Shanahan, Hugh and Straube, Armin and Eynden, Veerle Van den and Vandendorpe, Justine and Venkataram, Shanmugasundaram and Vieira, André and Wiljes, Cord and Wuttke, Ulrike and Yeomans, Joanne and Zhou, Biru},
	year = {2022},
	note = {Accepted: 2022-05-13T08:33:14Z
Artwork Medium: Print
Interview Medium: Print},
}

Softcover, 17x24

@misc{wienrich_ai_2022,
	title = {{AI} {Literacy}: {Kompetenzdimensionen} und {Einflussfaktoren} im {Kontext} von {Arbeit}},
	url = {https://www.denkfabrik-bmas.de/fileadmin/Downloads/Publikationen/AI_Literacy_Kompetenzdimensionen_und_Einflussfaktoren_im_Kontext_von_Arbeit.pdf},
	author = {Wienrich, Carolin and Carolus, Astrid and Augustin, Yannik and Markus, André},
	year = {2022},
}

@incollection{egger_natural_2022,
	address = {Cham},
	series = {Tourism on the {Verge}},
	title = {Natural {Language} {Processing} ({NLP}): {An} {Introduction}},
	isbn = {978-3-030-88389-8},
	shorttitle = {Natural {Language} {Processing} ({NLP})},
	url = {https://doi.org/10.1007/978-3-030-88389-8_15},
	doi = {10.1007/978-3-030-88389-8_15},
	abstract = {With the increase in internet usage, the amount of available textual data has also continued to increase rapidly. In addition, the development of stronger computers has enabled the processing of data to become much easier. The tourism field has a strong potential to utilize such data available on the internet; yet, on the other hand, a high proportion of available data is unlabelled and unprocessed. In order to use them effectively, new methods and new approaches are needed. In this regard, the area of Natural Language Processing (NLP) helps researchers to utilize textual data and develop an understanding of text analysis. By using machine learning approaches, text mining potential can expand enormously, leading to deeper insights, a better understanding of social phenomena, and, thus, also a better basis for decision-making. As such, this chapter will provide the reader with the basics of NLP as well as present the text pre-processing procedure in detail.},
	language = {en},
	urldate = {2023-09-14},
	booktitle = {Applied {Data} {Science} in {Tourism}: {Interdisciplinary} {Approaches}, {Methodologies}, and {Applications}},
	publisher = {Springer International Publishing},
	author = {Egger, Roman and Gokce, Enes},
	editor = {Egger, Roman},
	year = {2022},
	keywords = {Feature extraction, NER, POS, Pre-processing, Text cleaning},
	pages = {307--334},
}

With the increase in internet usage, the amount of available textual data has also continued to increase rapidly. In addition, the development of stronger computers has enabled the processing of data to become much easier. The tourism field has a strong potential to utilize such data available on the internet; yet, on the other hand, a high proportion of available data is unlabelled and unprocessed. In order to use them effectively, new methods and new approaches are needed. In this regard, the area of Natural Language Processing (NLP) helps researchers to utilize textual data and develop an understanding of text analysis. By using machine learning approaches, text mining potential can expand enormously, leading to deeper insights, a better understanding of social phenomena, and, thus, also a better basis for decision-making. As such, this chapter will provide the reader with the basics of NLP as well as present the text pre-processing procedure in detail.

@inproceedings{peverelli_process_2022,
	address = {Antwerp},
	title = {The {Process} of {Imitatio} {Through} {Stylometric} {Analysis}: the {Case} of {Terence}’s {Eunuchus}},
	abstract = {The Early Modern Era is at the forefront of a widespread enthusiasm for Latin works: texts from classical antiquity are given new life, widely re-printed, studied and even repeatedly staged, in the case of dramas, throughout Europe. Also, new Latin comedies are again written in quantities never seen before (at least 10,000 works published 1500 to 1800 are known). The authors themselves, within the game of literary imitation (the process of imitatio), start to mimic the style of ancient authors, and Terence’s dramas in particular were considered the prime sources of reuse for many decades. Via a case study ”the reception of Terence’s Eunuchus in Early Modern literature”, we take a deep dive into the mechanisms of literary imitation. Our analysis is based on four comedy corpora in Latin, Italian, French and English, spanning roughly 3 centuries (1400-1700). To assess the problem of language shi昀琀 and multi-language intercorpora analysis, we base our experiments on translations of the Eunuchus, one for each sub-corpus. Through the use of tools drawn from the 昀椀eld of Stylometry, we address the topic of text reuse and textual similarities between Terence’s text and Early-Modern corpora to get a better grasp on the internal 昀氀uctuations of the imitation game between Early Modern and Classical authors.},
	language = {en},
	author = {Peverelli, Andrea and van Erp, Marieke and Bloemendal, Jan},
	year = {2022},
	pages = {337--354},
}

The Early Modern Era is at the forefront of a widespread enthusiasm for Latin works: texts from classical antiquity are given new life, widely re-printed, studied and even repeatedly staged, in the case of dramas, throughout Europe. Also, new Latin comedies are again written in quantities never seen before (at least 10,000 works published 1500 to 1800 are known). The authors themselves, within the game of literary imitation (the process of imitatio), start to mimic the style of ancient authors, and Terence’s dramas in particular were considered the prime sources of reuse for many decades. Via a case study ”the reception of Terence’s Eunuchus in Early Modern literature”, we take a deep dive into the mechanisms of literary imitation. Our analysis is based on four comedy corpora in Latin, Italian, French and English, spanning roughly 3 centuries (1400-1700). To assess the problem of language shi昀琀 and multi-language intercorpora analysis, we base our experiments on translations of the Eunuchus, one for each sub-corpus. Through the use of tools drawn from the 昀椀eld of Stylometry, we address the topic of text reuse and textual similarities between Terence’s text and Early-Modern corpora to get a better grasp on the internal 昀氀uctuations of the imitation game between Early Modern and Classical authors.

@misc{nagy_stylometric_2022,
	title = {Some {Stylometric} {Remarks} on {Ovid}'s {Heroides} and the {Epistula} {Sapphus}},
	url = {http://arxiv.org/abs/2202.11864},
	doi = {10.48550/arXiv.2202.11864},
	abstract = {This article aims to contribute to two well-worn areas of debate in classical Latin philology, relating to Ovid's Heroides. The first is the question of the authenticity (and, to a lesser extent the correct position) of the letter placed fifteenth by almost every editor -- the so-called Epistula Sapphus (henceforth ES). The secondary question, although perhaps now less fervently debated, is the authenticity of the 'Double Heroides', placed by those who accept them as letters 16-21. I employ a variety of methods drawn from the domain of computational stylometry to consider the poetics and the lexico-grammatical features of these elegiac poems in the broader context of a corpus of 'shorter' (from 20 to 546 lines) elegiac works from five authors (266 poems in all) comprising more or less all of the non-fragmentary classical corpus. Based on a variety of techniques, every measure gives clear indication that the poetic style of the Heroides is Ovidian, but distinctive; they can be accurately isolated from Ovid more broadly. The Single and Double Heroides split into two clear groups, with the ES grouped consistently with the single letters. Furthermore, by comparing the style of the letters with the 'early' (although there are complications in this label) works of the Amores and the late works of the Ex Ponto, the evidence supports sequential composition -- meaning that the ES is correctly placed -- and, further, supports the growing consensus that the double letters were composed significantly later, in exile.},
	urldate = {2023-08-26},
	publisher = {arXiv},
	author = {Nagy, Ben},
	month = feb,
	year = {2022},
	note = {arXiv:2202.11864 [cs]},
	keywords = {Computer Science - Computation and Language},
}

This article aims to contribute to two well-worn areas of debate in classical Latin philology, relating to Ovid's Heroides. The first is the question of the authenticity (and, to a lesser extent the correct position) of the letter placed fifteenth by almost every editor – the so-called Epistula Sapphus (henceforth ES). The secondary question, although perhaps now less fervently debated, is the authenticity of the 'Double Heroides', placed by those who accept them as letters 16-21. I employ a variety of methods drawn from the domain of computational stylometry to consider the poetics and the lexico-grammatical features of these elegiac poems in the broader context of a corpus of 'shorter' (from 20 to 546 lines) elegiac works from five authors (266 poems in all) comprising more or less all of the non-fragmentary classical corpus. Based on a variety of techniques, every measure gives clear indication that the poetic style of the Heroides is Ovidian, but distinctive; they can be accurately isolated from Ovid more broadly. The Single and Double Heroides split into two clear groups, with the ES grouped consistently with the single letters. Furthermore, by comparing the style of the letters with the 'early' (although there are complications in this label) works of the Amores and the late works of the Ex Ponto, the evidence supports sequential composition – meaning that the ES is correctly placed – and, further, supports the growing consensus that the double letters were composed significantly later, in exile.

@article{nagy_rhyme_2022,
	title = {Rhyme in classical {Latin} poetry: {Stylistic} or stochastic?},
	volume = {37},
	issn = {2055-7671},
	shorttitle = {Rhyme in classical {Latin} poetry},
	url = {https://doi.org/10.1093/llc/fqab105},
	doi = {10.1093/llc/fqab105},
	abstract = {This study offers the first broad quantitative analysis of the use of rhyme in classical Latin hexameter and elegiac verse. The data and tools developed for the analysis are released under a permissive open source license. These include software to create an accurate phonetic transcription of Latin verse from the Musisque Deoque corpus; a system for scoring rhyme via phonetic similarity; and a system for generating large amounts of metrically correct, stochastic Latin verse (useful for analysis baselines). Further to this, some initial analysis is performed: first via descriptive statistics and then with two unsupervised multivariate analyses using dimension reduction methods. The study examines nineteen works by twelve authors, comprising about 96,000 lines. First and foremost, the results suggest that rhyme was consciously used by classical authors, but to different extents and in different ways. There is a solid and detectable stylistic separation between the use of rhyme in elegy and epic, and possibly also between satire and the rest. Within genres, authors can be stylistically separated with a small set of features. On the negative side, it appears that the stylistic signal from rhyme is fairly faint, and so forensic analysis (e.g. for authorship attribution) is not presently recommended on texts that are shorter than several thousand lines.},
	number = {4},
	urldate = {2023-08-26},
	journal = {Digital Scholarship in the Humanities},
	author = {Nagy, Ben},
	month = dec,
	year = {2022},
	pages = {1097--1118},
}

This study offers the first broad quantitative analysis of the use of rhyme in classical Latin hexameter and elegiac verse. The data and tools developed for the analysis are released under a permissive open source license. These include software to create an accurate phonetic transcription of Latin verse from the Musisque Deoque corpus; a system for scoring rhyme via phonetic similarity; and a system for generating large amounts of metrically correct, stochastic Latin verse (useful for analysis baselines). Further to this, some initial analysis is performed: first via descriptive statistics and then with two unsupervised multivariate analyses using dimension reduction methods. The study examines nineteen works by twelve authors, comprising about 96,000 lines. First and foremost, the results suggest that rhyme was consciously used by classical authors, but to different extents and in different ways. There is a solid and detectable stylistic separation between the use of rhyme in elegy and epic, and possibly also between satire and the rest. Within genres, authors can be stylistically separated with a small set of features. On the negative side, it appears that the stylistic signal from rhyme is fairly faint, and so forensic analysis (e.g. for authorship attribution) is not presently recommended on texts that are shorter than several thousand lines.

@article{granger_jupyter_2021,
	title = {Jupyter: {Thinking} and {Storytelling} {With} {Code} and {Data}},
	volume = {23},
	doi = {10.1109/MCSE.2021.3059263},
	number = {2},
	journal = {Computing in Science \& Engineering},
	author = {Granger, Brian E. and Pérez, Fernando},
	year = {2021},
	keywords = {Data science, Machine learning, Meteorology, Open source software, Scientific computing},
	pages = {7--14},
}

@article{tan_nascent_2021,
	title = {The {Nascent} {Case} for {Adopting} {Jupyter} {Notebooks} as a {Pedagogical} {Tool} for {Interdisciplinary} {Humanities}, {Social} {Science}, and {Arts} {Education}},
	volume = {23},
	url = {https://api.semanticscholar.org/CorpusID:232413487},
	journal = {Computing in Science \& Engineering},
	author = {Tan, Chiin-Rui and Geva, Sharon Broude and Colbry, Dirk},
	year = {2021},
	pages = {107--113},
}

@incollection{schmidt_data_2021,
	address = {Wiesbaden},
	title = {Data {Literacy} als ein essenzieller {Skill} für das 21. {Jahrhundert}},
	isbn = {978-3-658-33403-1},
	url = {https://doi.org/10.1007/978-3-658-33403-1_2},
	doi = {10.1007/978-3-658-33403-1_2},
	abstract = {Die heutige Geschäftswelt wird bestimmt durch Daten. Während das Problem normalerweise nicht ein Mangel an Daten ist, ist es vielmehr die Unfähigkeit, mit Daten adäquat umzugehen und daraus zielführende Schlussfolgerungen abzuleiten. Dabei ist Datenkompetenz heute genauso wichtig wie Lesen und Schreiben und eine Schlüsselqualifikation der Vierten Industriellen Revolution. Obgleich den meisten Unternehmen die Bedeutung von Daten in einer immer digitaleren Geschäftswelt durchaus bewusst ist, gibt es dennoch immer noch große Wissenslücken. Dieser Beitrag soll den Begriff Data Literacy und die damit verbundenen Fähigkeiten komprimiert beleuchten, während darüber hinaus mögliche Ansätze zur Integration des Themenfelds in Lehre und Praxis vorgestellt werden.},
	language = {de},
	urldate = {2024-01-02},
	booktitle = {Data {Science}: {Konzepte}, {Erfahrungen}, {Fallstudien} und {Praxis}},
	publisher = {Springer Fachmedien},
	author = {Schmidt, Andreas and Neifer, Thomas and Haag, Benedikt},
	editor = {Frick, Detlev and Gadatsch, Andreas and Kaufmann, Jens and Lankes, Birgit and Quix, Christoph and Schmidt, Andreas and Schmitz, Uwe},
	year = {2021},
	pages = {27--40},
}

Die heutige Geschäftswelt wird bestimmt durch Daten. Während das Problem normalerweise nicht ein Mangel an Daten ist, ist es vielmehr die Unfähigkeit, mit Daten adäquat umzugehen und daraus zielführende Schlussfolgerungen abzuleiten. Dabei ist Datenkompetenz heute genauso wichtig wie Lesen und Schreiben und eine Schlüsselqualifikation der Vierten Industriellen Revolution. Obgleich den meisten Unternehmen die Bedeutung von Daten in einer immer digitaleren Geschäftswelt durchaus bewusst ist, gibt es dennoch immer noch große Wissenslücken. Dieser Beitrag soll den Begriff Data Literacy und die damit verbundenen Fähigkeiten komprimiert beleuchten, während darüber hinaus mögliche Ansätze zur Integration des Themenfelds in Lehre und Praxis vorgestellt werden.

@incollection{borek_information_2021,
	address = {Glückstadt},
	series = {Schriften zur {Informationswissenschaft}},
	title = {Information {Organization} and {Access} in {Digital} {Humanities}: {TaDiRAH} {Revised}, {Formalized} and {FAIR}},
	volume = {74},
	isbn = {978-3-86488-172-5},
	shorttitle = {Information {Organization} and {Access} in {Digital} {Humanities}},
	url = {https://epub.uni-regensburg.de/44951/},
	abstract = {Classifying and categorizing the activities that comprise the digital humanities (DH) has been a longstanding area of interest for many practitioners in this field, fueled by ongoing attempts to define the field both within the academic and public sphere. Several European initiatives are currently shaping advanced research infrastructures that would benefit from an implementation of a suiting taxonomy. Therefore, new humanities and information science collaborations have been formed to provide a service that meets their needs. This working paper presents the transformation of the Taxonomy of Digital Research Activities in the Humanities (TaDiRAH) in order to make it machine-readable and become a formalized taxonomy. This includes the methodology and realization containing a complete revision of the original version, decisions in modelling, the implementation as well as organization of ongoing and future tasks. TaDiRAH addresses a wide range of humanities disciplines and integrates application areas from philologies as well as epigraphy, and musicology to name just a few. For this reason, the decision in favor of SKOS was made purely pragmatically in terms of technology, concept and domains. New language versions can now be easily integrated and low-threshold term extensions can be carried out via Wikidata. The new TaDiRAH not only represents a knowledge organization system ( KOS ) which has recently been released as version 2.0. According to the FAIR principles this new version improves the Findability, Accessibility, Interoperability, and Reuse of research data and digital assets in the digital humanities.},
	language = {en},
	urldate = {2025-07-06},
	booktitle = {Information between {Data} and {Knowledge}},
	publisher = {Werner Hülsbusch},
	author = {Borek, Luise and Hastik, Canan and Khramova, Vera and Illmayer, Klaus and Geiger, Jonathan D.},
	year = {2021},
	pages = {321--332},
}

Classifying and categorizing the activities that comprise the digital humanities (DH) has been a longstanding area of interest for many practitioners in this field, fueled by ongoing attempts to define the field both within the academic and public sphere. Several European initiatives are currently shaping advanced research infrastructures that would benefit from an implementation of a suiting taxonomy. Therefore, new humanities and information science collaborations have been formed to provide a service that meets their needs. This working paper presents the transformation of the Taxonomy of Digital Research Activities in the Humanities (TaDiRAH) in order to make it machine-readable and become a formalized taxonomy. This includes the methodology and realization containing a complete revision of the original version, decisions in modelling, the implementation as well as organization of ongoing and future tasks. TaDiRAH addresses a wide range of humanities disciplines and integrates application areas from philologies as well as epigraphy, and musicology to name just a few. For this reason, the decision in favor of SKOS was made purely pragmatically in terms of technology, concept and domains. New language versions can now be easily integrated and low-threshold term extensions can be carried out via Wikidata. The new TaDiRAH not only represents a knowledge organization system ( KOS ) which has recently been released as version 2.0. According to the FAIR principles this new version improves the Findability, Accessibility, Interoperability, and Reuse of research data and digital assets in the digital humanities.

@article{ng_conceptualizing_2021,
	title = {Conceptualizing {AI} literacy: {An} exploratory review},
	volume = {2},
	issn = {2666-920X},
	shorttitle = {Conceptualizing {AI} literacy},
	url = {https://www.sciencedirect.com/science/article/pii/S2666920X21000357},
	doi = {10.1016/j.caeai.2021.100041},
	abstract = {Artificial Intelligence (AI) has spread across industries (e.g., business, science, art, education) to enhance user experience, improve work efficiency, and create many future job opportunities. However, public understanding of AI technologies and how to define AI literacy is under-explored. This vision poses upcoming challenges for our next generation to learn about AI. On this note, an exploratory review was conducted to conceptualize the newly emerging concept “AI literacy”, in search for a sound theoretical foundation to define, teach and evaluate AI literacy. Grounded in literature on 30 existing peer-reviewed articles, this review proposed four aspects (i.e., know and understand, use and apply, evaluate and create, and ethical issues) for fostering AI literacy based on the adaptation of classic literacies. This study sheds light on the consolidated definition, teaching, and ethical concerns on AI literacy, establishing the groundwork for future research such as competency development and assessment criteria on AI literacy.},
	urldate = {2023-12-21},
	journal = {Computers and Education: Artificial Intelligence},
	author = {Ng, Davy Tsz Kit and Leung, Jac Ka Lok and Chu, Samuel Kai Wah and Qiao, Maggie Shen},
	month = jan,
	year = {2021},
	keywords = {AI ethics, AI in education, AI learning and teaching, AI literacy, AI literacy questionnaire},
	pages = {100041},
}

Artificial Intelligence (AI) has spread across industries (e.g., business, science, art, education) to enhance user experience, improve work efficiency, and create many future job opportunities. However, public understanding of AI technologies and how to define AI literacy is under-explored. This vision poses upcoming challenges for our next generation to learn about AI. On this note, an exploratory review was conducted to conceptualize the newly emerging concept “AI literacy”, in search for a sound theoretical foundation to define, teach and evaluate AI literacy. Grounded in literature on 30 existing peer-reviewed articles, this review proposed four aspects (i.e., know and understand, use and apply, evaluate and create, and ethical issues) for fostering AI literacy based on the adaptation of classic literacies. This study sheds light on the consolidated definition, teaching, and ethical concerns on AI literacy, establishing the groundwork for future research such as competency development and assessment criteria on AI literacy.

@article{ng_ai_2021,
	title = {{AI} {Literacy}: {Definition}, {Teaching}, {Evaluation} and {Ethical} {Issues}},
	volume = {58},
	issn = {2373-9231},
	shorttitle = {{AI} {Literacy}},
	url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/pra2.487},
	doi = {10.1002/pra2.487},
	abstract = {Artificial Intelligence (AI) is at the top of the agenda for education leaders today in educating the next generation across the globe. However, public understanding of AI technologies and how to define AI literacy is under-explored. This vision poses upcoming challenges for our next generation to learn about AI. On this note, an exploratory review was conducted to conceptualize the newly emerging concept “AI literacy”, in search for a sound theoretical foundation to define, teach and evaluate AI literacy. Grounded in literature on 18 existing peer-reviewed articles, this review proposed four aspects (i.e, know and understand, use, evaluate, and ethical issues) for fostering AI literacy based on the adaptation of classic literacies. This study sheds light on the consolidated definition, teaching, and ethical concerns on AI literacy, establishing the groundwork for future research such as competency development and assessment criteria on AI literacy.},
	language = {en},
	number = {1},
	urldate = {2023-12-21},
	journal = {Proceedings of the Association for Information Science and Technology},
	author = {Ng, Davy Tsz Kit and Leung, Jac Ka Lok and Chu, Kai Wah Samuel and Qiao, Maggie Shen},
	year = {2021},
	note = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/pra2.487},
	keywords = {AI ethics, AI in education, AI learning and teaching, AI literacy, AI literacy questionnaire},
	pages = {504--509},
}

Artificial Intelligence (AI) is at the top of the agenda for education leaders today in educating the next generation across the globe. However, public understanding of AI technologies and how to define AI literacy is under-explored. This vision poses upcoming challenges for our next generation to learn about AI. On this note, an exploratory review was conducted to conceptualize the newly emerging concept “AI literacy”, in search for a sound theoretical foundation to define, teach and evaluate AI literacy. Grounded in literature on 18 existing peer-reviewed articles, this review proposed four aspects (i.e, know and understand, use, evaluate, and ethical issues) for fostering AI literacy based on the adaptation of classic literacies. This study sheds light on the consolidated definition, teaching, and ethical concerns on AI literacy, establishing the groundwork for future research such as competency development and assessment criteria on AI literacy.

@techreport{gledic_survey_2021,
	title = {Survey of curricula: {Linguistics} and language-related degrees in {Europe}},
	shorttitle = {Survey of curricula},
	url = {https://zenodo.org/record/5030861},
	doi = {10.5281/zenodo.5030860},
	abstract = {The needs analysis of the UPSKILLS project is the foundation for all subsequent project activities, and the survey of curricula as its first step is designed to provide insights for finetuning the interventions and materials that will be designed during the lifetime of the project, as well as for enlarging the pool of stakeholders to whom the project results will be disseminated. The survey of curricula has several steps: drawing a list of European language and linguistics degrees from international ranking websites, selecting and analyzing a representative sample of degrees based on a set of indicators agreed upon by all partners, and additional studying of a selection of degrees that the partners identified as exemplary in the context of the UPSKILLS project.},
	urldate = {2023-10-02},
	institution = {Zenodo},
	author = {Gledić, Jelena and Đukanović, Maja and Miličević Petrović, Maja and van der Lek, Iulianna and Assimakopoulos, Stavros},
	month = jun,
	year = {2021},
}

The needs analysis of the UPSKILLS project is the foundation for all subsequent project activities, and the survey of curricula as its first step is designed to provide insights for finetuning the interventions and materials that will be designed during the lifetime of the project, as well as for enlarging the pool of stakeholders to whom the project results will be disseminated. The survey of curricula has several steps: drawing a list of European language and linguistics degrees from international ranking websites, selecting and analyzing a representative sample of degrees based on a set of indicators agreed upon by all partners, and additional studying of a selection of degrees that the partners identified as exemplary in the context of the UPSKILLS project.

@article{gebru_datasheets_2021,
	title = {Datasheets for {Datasets}},
	volume = {64},
	url = {https://dl.acm.org/doi/10.1145/3458723},
	abstract = {The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains. To address this gap, we propose datasheets for datasets. In the electronics industry, every component, no matter how simple or complex, is accompanied with a datasheet that describes its operating characteristics, test results, recommended uses, and other information. By analogy, we propose that every dataset be accompanied with a datasheet that documents its motivation, composition, collection process, recommended uses, and so on. Datasheets for datasets will facilitate better communication between dataset creators and dataset consumers, and encourage the machine learning community to prioritize transparency and accountability.},
	number = {12},
	urldate = {2021-12-03},
	journal = {Communications of the ACM},
	author = {Gebru, Timnit and Morgenstern, Jamie and Vecchione, Briana and Vaughan, Jennifer Wortman and Wallach, Hanna and Daumé III, Hal and Crawford, Kate},
	year = {2021},
	note = {arXiv: 1803.09010},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Databases, Computer Science - Machine Learning},
	pages = {86--92},
}

The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains. To address this gap, we propose datasheets for datasets. In the electronics industry, every component, no matter how simple or complex, is accompanied with a datasheet that describes its operating characteristics, test results, recommended uses, and other information. By analogy, we propose that every dataset be accompanied with a datasheet that documents its motivation, composition, collection process, recommended uses, and so on. Datasheets for datasets will facilitate better communication between dataset creators and dataset consumers, and encourage the machine learning community to prioritize transparency and accountability.

@misc{weidinger_ethical_2021,
	title = {Ethical and social risks of harm from {Language} {Models}},
	url = {http://arxiv.org/abs/2112.04359},
	abstract = {This paper aims to help structure the risk landscape associated with large-scale Language Models (LMs). In order to foster advances in responsible innovation, an in-depth understanding of the potential risks posed by these models is needed. A wide range of established and anticipated risks are analysed in detail, drawing on multidisciplinary expertise and literature from computer science, linguistics, and social sciences. We outline six specific risk areas: I. Discrimination, Exclusion and Toxicity, II. Information Hazards, III. Misinformation Harms, V. Malicious Uses, V. Human-Computer Interaction Harms, VI. Automation, Access, and Environmental Harms. The first area concerns the perpetuation of stereotypes, unfair discrimination, exclusionary norms, toxic language, and lower performance by social group for LMs. The second focuses on risks from private data leaks or LMs correctly inferring sensitive information. The third addresses risks arising from poor, false or misleading information including in sensitive domains, and knock-on risks such as the erosion of trust in shared information. The fourth considers risks from actors who try to use LMs to cause harm. The fifth focuses on risks specific to LLMs used to underpin conversational agents that interact with human users, including unsafe use, manipulation or deception. The sixth discusses the risk of environmental harm, job automation, and other challenges that may have a disparate effect on different social groups or communities. In total, we review 21 risks in-depth. We discuss the points of origin of different risks and point to potential mitigation approaches. Lastly, we discuss organisational responsibilities in implementing mitigations, and the role of collaboration and participation. We highlight directions for further research, particularly on expanding the toolkit for assessing and evaluating the outlined risks in LMs.},
	urldate = {2022-03-22},
	publisher = {arXiv},
	author = {Weidinger, Laura and Mellor, John and Rauh, Maribeth and Griffin, Conor and Uesato, Jonathan and Huang, Po-Sen and Cheng, Myra and Glaese, Mia and Balle, Borja and Kasirzadeh, Atoosa and Kenton, Zac and Brown, Sasha and Hawkins, Will and Stepleton, Tom and Biles, Courtney and Birhane, Abeba and Haas, Julia and Rimell, Laura and Hendricks, Lisa Anne and Isaac, William and Legassick, Sean and Irving, Geoffrey and Gabriel, Iason},
	month = dec,
	year = {2021},
	note = {arXiv: 2112.04359},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Computers and Society},
}

This paper aims to help structure the risk landscape associated with large-scale Language Models (LMs). In order to foster advances in responsible innovation, an in-depth understanding of the potential risks posed by these models is needed. A wide range of established and anticipated risks are analysed in detail, drawing on multidisciplinary expertise and literature from computer science, linguistics, and social sciences. We outline six specific risk areas: I. Discrimination, Exclusion and Toxicity, II. Information Hazards, III. Misinformation Harms, V. Malicious Uses, V. Human-Computer Interaction Harms, VI. Automation, Access, and Environmental Harms. The first area concerns the perpetuation of stereotypes, unfair discrimination, exclusionary norms, toxic language, and lower performance by social group for LMs. The second focuses on risks from private data leaks or LMs correctly inferring sensitive information. The third addresses risks arising from poor, false or misleading information including in sensitive domains, and knock-on risks such as the erosion of trust in shared information. The fourth considers risks from actors who try to use LMs to cause harm. The fifth focuses on risks specific to LLMs used to underpin conversational agents that interact with human users, including unsafe use, manipulation or deception. The sixth discusses the risk of environmental harm, job automation, and other challenges that may have a disparate effect on different social groups or communities. In total, we review 21 risks in-depth. We discuss the points of origin of different risks and point to potential mitigation approaches. Lastly, we discuss organisational responsibilities in implementing mitigations, and the role of collaboration and participation. We highlight directions for further research, particularly on expanding the toolkit for assessing and evaluating the outlined risks in LMs.

@misc{lin_truthfulqa_2021,
	title = {Truthfulqa: {Measuring} how models mimic human falsehoods},
	url = {https://arxiv.org/pdf/2109.07958.pdf},
	publisher = {arXiv},
	author = {Lin, Stephanie and Hilton, Jacob and Evans, Owain},
	year = {2021},
}

@techreport{hellman_big_2021,
	address = {Rochester, NY},
	type = {{SSRN} {Scholarly} {Paper}},
	title = {Big {Data} and {Compounding} {Injustice}},
	url = {https://papers.ssrn.com/abstract=3840175},
	abstract = {This article argues that the fact that an action will compound a prior injustice counts as a reason against doing the action.  I call this reason The Anti-compounding Injustice Principle or ACI.  Compounding injustice and the ACI principle are likely to be relevant when analyzing the moral issues raised by “big data” and its combination with the computational power of machine learning and artificial intelligence.  Past injustice can infect the data used in algorithmic decisions in two distinct ways.  Sometimes prior injustice undermines the accuracy of the data itself.  In these contexts, improving accuracy will also help to avoid compounding injustice.  Other times, past injustice produces real world differences among people with regard to skills, health, wealth, and other traits that employers, lenders and others seek to measure.  When decisions are based on accurate data that itself results from prior injustice, these decisions can also compound injustice.  This second dynamic has received less attention than the first but is especially important because improving the accuracy of data will not mitigate this unfairness.},
	language = {en},
	number = {3840175},
	urldate = {2022-05-18},
	institution = {Social Science Research Network},
	author = {Hellman, Deborah},
	month = may,
	year = {2021},
	keywords = {Algorithms, Artificial intelligence, Discrimination, accuracy, big data, disparate impact, ethics, injustice},
}

This article argues that the fact that an action will compound a prior injustice counts as a reason against doing the action. I call this reason The Anti-compounding Injustice Principle or ACI. Compounding injustice and the ACI principle are likely to be relevant when analyzing the moral issues raised by “big data” and its combination with the computational power of machine learning and artificial intelligence. Past injustice can infect the data used in algorithmic decisions in two distinct ways. Sometimes prior injustice undermines the accuracy of the data itself. In these contexts, improving accuracy will also help to avoid compounding injustice. Other times, past injustice produces real world differences among people with regard to skills, health, wealth, and other traits that employers, lenders and others seek to measure. When decisions are based on accurate data that itself results from prior injustice, these decisions can also compound injustice. This second dynamic has received less attention than the first but is especially important because improving the accuracy of data will not mitigate this unfairness.

@inproceedings{keersmaekers_glaux_2021,
	title = {The {GLAUx} corpus: methodological issues in designing a long-term, diverse, multi-layered corpus of {Ancient} {Greek}},
	url = {https://aclanthology.org/2021.lchange-1.6.pdf},
	author = {Keersmaekers, Alek},
	year = {2021},
	pages = {39--50},
}

@article{heath_applied_2021,
	series = {{ISAW} {Papers}},
	title = {Applied {Use} of {JSON}, {GeoJSON}, {JSON}-{LD}, {SPARQL}, and {IPython} {Notebooks} for {Representing} and {Interacting} with {Small} {Datasets}},
	volume = {20},
	url = {https://dlib.nyu.edu/awdl/isaw/isaw-papers/20-13/},
	journal = {Linked Open Data for the Ancient Mediterranean: Structures, Practices, Prospects},
	author = {Heath, Sebastian},
	year = {2021},
}

@article{beyer_spracherwerb_2021,
	title = {Spracherwerb in der {Forschung} - neue {Möglichkeiten} dank {Digitalisierung}},
	volume = {64},
	issn = {1432-7511},
	url = {https://journals.ub.uni-heidelberg.de/index.php/fc/article/view/81215},
	doi = {10.11588/fc.2021.1.81215},
	number = {1},
	journal = {Forum Classicum},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = mar,
	year = {2021},
	pages = {4--10},
}

@techreport{schmid_kibildung_2021,
	type = {Schlussbericht},
	title = {{KI}@{Bildung}: {Lehren} und {Lernen} in der {Schule} mit {Werkzeugen} {Künstlicher} {Intelligenz}},
	url = {https://www.telekom-stiftung.de/sites/default/files/files/media/publications/KI%20Bildung%20Schlussbericht.pdf},
	institution = {Deutsche Telekom Stiftung},
	author = {Schmid, Ulrich and Blanc, Berit and Toepel, Michael and Pinkwart, Niels and Drachsler, Hendrik},
	year = {2021},
	note = {mmb Institut – Gesellschaft für Medien- und Kompetenzforschung mbH},
	pages = {1--58},
}

@article{chastang_named_2021,
	title = {A {Named} {Entity} {Recognition} {Model} for {Medieval} {Latin} {Charters}},
	volume = {15},
	issn = {1938-4122},
	url = {https://orbilu.uni.lu/bitstream/10993/59645/1/000574.pdf},
	number = {4},
	journal = {Digital Humanities Quarterly},
	author = {Chastang, Pierre and Torres Aguilar, Sergio Octavio and Tannier, Xavier},
	year = {2021},
}

@article{torregrossa_survey_2021,
	title = {A survey on training and evaluation of word embeddings.},
	doi = {https://doi.org/10.1007/s41060-021-00242-8},
	abstract = {Word embeddings have proven to be effective for many natural language processing tasks by providing word representations integrating prior knowledge. In this article, we focus on the algorithms and models used to compute those representations and on their methods of evaluation. Many new techniques were developed in a short amount of time, and there is no unified terminology to emphasise strengths and weaknesses of those methods. Based on the state of the art, we propose a thorough terminology to help with the classification of these various models and their evaluations. We also provide comparisons of those algorithms and methods, highlighting open problems and research paths, as well as a compilation of popular evaluation metrics and datasets. This survey gives: (1) an exhaustive description and terminology of currently investigated word embeddings, (2) a clear segmentation of evaluation methods and their associated datasets, and (3) high-level properties to indicate pros and cons of each solution.},
	number = {11},
	journal = {Int J Data Sci Anal},
	author = {Torregrossa, F. and Allesiardo, R. and Claveau, V.},
	year = {2021},
	keywords = {Word Embeddings},
	pages = {85--103},
}

Word embeddings have proven to be effective for many natural language processing tasks by providing word representations integrating prior knowledge. In this article, we focus on the algorithms and models used to compute those representations and on their methods of evaluation. Many new techniques were developed in a short amount of time, and there is no unified terminology to emphasise strengths and weaknesses of those methods. Based on the state of the art, we propose a thorough terminology to help with the classification of these various models and their evaluations. We also provide comparisons of those algorithms and methods, highlighting open problems and research paths, as well as a compilation of popular evaluation metrics and datasets. This survey gives: (1) an exhaustive description and terminology of currently investigated word embeddings, (2) a clear segmentation of evaluation methods and their associated datasets, and (3) high-level properties to indicate pros and cons of each solution.

@misc{adwait_beginners_2021,
	title = {A {Beginner}'s {Introduction} to {NER} ({Named} {Entity} {Recognition})},
	url = {https://www.analyticsvidhya.com/blog/2021/11/a-beginners-introduction-to-ner-named-entity-recognition/},
	abstract = {NER(Named Entity recognition), a popular method that is used for recognizing entities that are present in a text document.},
	language = {en},
	urldate = {2025-07-07},
	journal = {Analytics Vidhya},
	author = {Adwait},
	month = nov,
	year = {2021},
}

NER(Named Entity recognition), a popular method that is used for recognizing entities that are present in a text document.

@article{ehrmann_named_2021,
	title = {Named entity recognition and classification in historical documents: {A} survey},
	issn = {0360-0300},
	url = {https://dl.acm.org/doi/pdf/10.1145/3604931},
	journal = {ACM Computing Surveys},
	publisher = {ACM New York, NY},
	author = {Ehrmann, Maud and Hamdi, Ahmed and Pontes, Elvys Linhares and Romanello, Matteo and Doucet, Antoine},
	year = {2021},
}

@article{ehrmann_named_2021,
	title = {Named {Entity} {Recognition} and {Classification} on {Historical} {Documents}: {A} {Survey}},
	volume = {56},
	issn = {0360-0300, 1557-7341},
	shorttitle = {Named {Entity} {Recognition} and {Classification} on {Historical} {Documents}},
	url = {http://arxiv.org/abs/2109.11406},
	doi = {10.1145/3604931},
	abstract = {After decades of massive digitisation, an unprecedented amount of historical documents is available in digital format, along with their machine-readable texts. While this represents a major step forward with respect to preservation and accessibility, it also opens up new opportunities in terms of content mining and the next fundamental challenge is to develop appropriate technologies to efficiently search, retrieve and explore information from this 'big data of the past'. Among semantic indexing opportunities, the recognition and classification of named entities are in great demand among humanities scholars. Yet, named entity recognition (NER) systems are heavily challenged with diverse, historical and noisy inputs. In this survey, we present the array of challenges posed by historical documents to NER, inventory existing resources, describe the main approaches deployed so far, and identify key priorities for future developments.},
	number = {2},
	urldate = {2025-01-01},
	journal = {ACM Computing Surveys},
	author = {Ehrmann, Maud and Hamdi, Ahmed and Pontes, Elvys Linhares and Romanello, Matteo and Doucet, Antoine},
	month = sep,
	year = {2021},
	note = {arXiv:2109.11406 [cs]},
	keywords = {Computer Science - Computation and Language, Computer Science - Machine Learning},
	pages = {1--47},
}

After decades of massive digitisation, an unprecedented amount of historical documents is available in digital format, along with their machine-readable texts. While this represents a major step forward with respect to preservation and accessibility, it also opens up new opportunities in terms of content mining and the next fundamental challenge is to develop appropriate technologies to efficiently search, retrieve and explore information from this 'big data of the past'. Among semantic indexing opportunities, the recognition and classification of named entities are in great demand among humanities scholars. Yet, named entity recognition (NER) systems are heavily challenged with diverse, historical and noisy inputs. In this survey, we present the array of challenges posed by historical documents to NER, inventory existing resources, describe the main approaches deployed so far, and identify key priorities for future developments.

@inproceedings{gehrmann_gem_2021,
	address = {Online},
	title = {The {GEM} {Benchmark}: {Natural} {Language} {Generation}, its {Evaluation} and {Metrics}},
	shorttitle = {The {GEM} {Benchmark}},
	url = {https://aclanthology.org/2021.gem-1.10/},
	doi = {10.18653/v1/2021.gem-1.10},
	abstract = {We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this moving target, new models often still evaluate on divergent anglo-centric corpora with well-established, but flawed, metrics. This disconnect makes it challenging to identify the limitations of current models and opportunities for progress. Addressing this limitation, GEM provides an environment in which models can easily be applied to a wide set of tasks and in which evaluation strategies can be tested. Regular updates to the benchmark will help NLG research become more multilingual and evolve the challenge alongside models. This paper serves as the description of the data for the 2021 shared task at the associated GEM Workshop.},
	urldate = {2025-01-15},
	booktitle = {Proceedings of the 1st {Workshop} on {Natural} {Language} {Generation}, {Evaluation}, and {Metrics} ({GEM} 2021)},
	publisher = {Association for Computational Linguistics},
	author = {Gehrmann, Sebastian and Adewumi, Tosin and Aggarwal, Karmanya and Ammanamanchi, Pawan Sasanka and Aremu, Anuoluwapo and Bosselut, Antoine and Chandu, Khyathi Raghavi and Clinciu, Miruna-Adriana and Das, Dipanjan and Dhole, Kaustubh and Du, Wanyu and Durmus, Esin and Dušek, Ondřej and Emezue, Chris Chinenye and Gangal, Varun and Garbacea, Cristina and Hashimoto, Tatsunori and Hou, Yufang and Jernite, Yacine and Jhamtani, Harsh and Ji, Yangfeng and Jolly, Shailza and Kale, Mihir and Kumar, Dhruv and Ladhak, Faisal and Madaan, Aman and Maddela, Mounica and Mahajan, Khyati and Mahamood, Saad and Majumder, Bodhisattwa Prasad and Martins, Pedro Henrique and McMillan-Major, Angelina and Mille, Simon and van Miltenburg, Emiel and Nadeem, Moin and Narayan, Shashi and Nikolaev, Vitaly and Niyongabo Rubungo, Andre and Osei, Salomey and Parikh, Ankur and Perez-Beltrachini, Laura and Rao, Niranjan Ramesh and Raunak, Vikas and Rodriguez, Juan Diego and Santhanam, Sashank and Sedoc, João and Sellam, Thibault and Shaikh, Samira and Shimorina, Anastasia and Sobrevilla Cabezudo, Marco Antonio and Strobelt, Hendrik and Subramani, Nishant and Xu, Wei and Yang, Diyi and Yerukola, Akhila and Zhou, Jiawei},
	editor = {Bosselut, Antoine and Durmus, Esin and Gangal, Varun Prashant and Gehrmann, Sebastian and Jernite, Yacine and Perez-Beltrachini, Laura and Shaikh, Samira and Xu, Wei},
	month = aug,
	year = {2021},
	pages = {96--120},
}

We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this moving target, new models often still evaluate on divergent anglo-centric corpora with well-established, but flawed, metrics. This disconnect makes it challenging to identify the limitations of current models and opportunities for progress. Addressing this limitation, GEM provides an environment in which models can easily be applied to a wide set of tasks and in which evaluation strategies can be tested. Regular updates to the benchmark will help NLG research become more multilingual and evolve the challenge alongside models. This paper serves as the description of the data for the 2021 shared task at the associated GEM Workshop.

@article{nasar_named_2021,
	title = {Named {Entity} {Recognition} and {Relation} {Extraction}: {State}-of-the-{Art}},
	volume = {54},
	issn = {0360-0300},
	shorttitle = {Named {Entity} {Recognition} and {Relation} {Extraction}},
	url = {https://doi.org/10.1145/3445965},
	doi = {10.1145/3445965},
	abstract = {With the advent of Web 2.0, there exist many online platforms that result in massive textual-data production. With ever-increasing textual data at hand, it is of immense importance to extract information nuggets from this data. One approach towards effective harnessing of this unstructured textual data could be its transformation into structured text. Hence, this study aims to present an overview of approaches that can be applied to extract key insights from textual data in a structured way. For this, Named Entity Recognition and Relation Extraction are being majorly addressed in this review study. The former deals with identification of named entities, and the latter deals with problem of extracting relation between set of entities. This study covers early approaches as well as the developments made up till now using machine learning models. Survey findings conclude that deep-learning-based hybrid and joint models are currently governing the state-of-the-art. It is also observed that annotated benchmark datasets for various textual-data generators such as Twitter and other social forums are not available. This scarcity of dataset has resulted into relatively less progress in these domains. Additionally, the majority of the state-of-the-art techniques are offline and computationally expensive. Last, with increasing focus on deep-learning frameworks, there is need to understand and explain the under-going processes in deep architectures.},
	number = {1},
	urldate = {2025-01-01},
	journal = {ACM Comput. Surv.},
	author = {Nasar, Zara and Jaffry, Syed Waqar and Malik, Muhammad Kamran},
	month = feb,
	year = {2021},
	pages = {20:1--20:39},
}

With the advent of Web 2.0, there exist many online platforms that result in massive textual-data production. With ever-increasing textual data at hand, it is of immense importance to extract information nuggets from this data. One approach towards effective harnessing of this unstructured textual data could be its transformation into structured text. Hence, this study aims to present an overview of approaches that can be applied to extract key insights from textual data in a structured way. For this, Named Entity Recognition and Relation Extraction are being majorly addressed in this review study. The former deals with identification of named entities, and the latter deals with problem of extracting relation between set of entities. This study covers early approaches as well as the developments made up till now using machine learning models. Survey findings conclude that deep-learning-based hybrid and joint models are currently governing the state-of-the-art. It is also observed that annotated benchmark datasets for various textual-data generators such as Twitter and other social forums are not available. This scarcity of dataset has resulted into relatively less progress in these domains. Additionally, the majority of the state-of-the-art techniques are offline and computationally expensive. Last, with increasing focus on deep-learning frameworks, there is need to understand and explain the under-going processes in deep architectures.

@misc{gebru_datasheets_2021,
	title = {Datasheets for {Datasets}},
	url = {http://arxiv.org/abs/1803.09010},
	doi = {10.48550/arXiv.1803.09010},
	abstract = {The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains. To address this gap, we propose datasheets for datasets. In the electronics industry, every component, no matter how simple or complex, is accompanied with a datasheet that describes its operating characteristics, test results, recommended uses, and other information. By analogy, we propose that every dataset be accompanied with a datasheet that documents its motivation, composition, collection process, recommended uses, and so on. Datasheets for datasets will facilitate better communication between dataset creators and dataset consumers, and encourage the machine learning community to prioritize transparency and accountability.},
	urldate = {2024-09-03},
	publisher = {arXiv},
	author = {Gebru, Timnit and Morgenstern, Jamie and Vecchione, Briana and Vaughan, Jennifer Wortman and Wallach, Hanna and Daumé III, Hal and Crawford, Kate},
	month = dec,
	year = {2021},
	note = {arXiv:1803.09010 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Databases, Computer Science - Machine Learning},
}

The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains. To address this gap, we propose datasheets for datasets. In the electronics industry, every component, no matter how simple or complex, is accompanied with a datasheet that describes its operating characteristics, test results, recommended uses, and other information. By analogy, we propose that every dataset be accompanied with a datasheet that documents its motivation, composition, collection process, recommended uses, and so on. Datasheets for datasets will facilitate better communication between dataset creators and dataset consumers, and encourage the machine learning community to prioritize transparency and accountability.

@inproceedings{singh_pilot_2021,
	address = {Punta Cana, Dominican Republic (online)},
	title = {A {Pilot} {Study} for {BERT} {Language} {Modelling} and {Morphological} {Analysis} for {Ancient} and {Medieval} {Greek}},
	url = {https://aclanthology.org/2021.latechclfl-1.15},
	doi = {10.18653/v1/2021.latechclfl-1.15},
	abstract = {This paper presents a pilot study to automatic linguistic preprocessing of Ancient and Byzantine Greek, and morphological analysis more specifically. To this end, a novel subword-based BERT language model was trained on the basis of a varied corpus of Modern, Ancient and Post-classical Greek texts. Consequently, the obtained BERT embeddings were incorporated to train a fine-grained Part-of-Speech tagger for Ancient and Byzantine Greek. In addition, a corpus of Greek Epigrams was manually annotated and the resulting gold standard was used to evaluate the performance of the morphological analyser on Byzantine Greek. The experimental results show very good perplexity scores (4.9) for the BERT language model and state-of-the-art performance for the fine-grained Part-of-Speech tagger for in-domain data (treebanks containing a mixture of Classical and Medieval Greek), as well as for the newly created Byzantine Greek gold standard data set. The language models and associated code are made available for use at https://github.com/pranaydeeps/Ancient-Greek-BERT},
	urldate = {2023-10-05},
	booktitle = {Proceedings of the 5th {Joint} {SIGHUM} {Workshop} on {Computational} {Linguistics} for {Cultural} {Heritage}, {Social} {Sciences}, {Humanities} and {Literature}},
	publisher = {Association for Computational Linguistics},
	author = {Singh, Pranaydeep and Rutten, Gorik and Lefever, Els},
	month = nov,
	year = {2021},
	pages = {128--137},
}

This paper presents a pilot study to automatic linguistic preprocessing of Ancient and Byzantine Greek, and morphological analysis more specifically. To this end, a novel subword-based BERT language model was trained on the basis of a varied corpus of Modern, Ancient and Post-classical Greek texts. Consequently, the obtained BERT embeddings were incorporated to train a fine-grained Part-of-Speech tagger for Ancient and Byzantine Greek. In addition, a corpus of Greek Epigrams was manually annotated and the resulting gold standard was used to evaluate the performance of the morphological analyser on Byzantine Greek. The experimental results show very good perplexity scores (4.9) for the BERT language model and state-of-the-art performance for the fine-grained Part-of-Speech tagger for in-domain data (treebanks containing a mixture of Classical and Medieval Greek), as well as for the newly created Byzantine Greek gold standard data set. The language models and associated code are made available for use at https://github.com/pranaydeeps/Ancient-Greek-BERT

@misc{sprugnoli_sentiment_2021,
	title = {Sentiment {Analysis} for {Latin}: a {Journey} from {Seneca} to {Thomas} {Aquinas}},
	shorttitle = {Sentiment {Analysis} for {Latin}},
	url = {https://zenodo.org/record/4575431#.YKelWoMzbJw},
	doi = {10.5281/zenodo.4575431},
	abstract = {While the main applications of resources and tools for sentiment analysis typically fall within the scope of fields like customer experience and social media monitoring, there is an increasing interest in extending their range to texts written in ancient and historical languages. Such interest mirrors the substantial growth of the area dedicated to building and using linguistic resources for these languages, which are essential for accessing and understanding the Classical tradition. In this talk, we will present the methodology we followed to create and evaluate a new set of Latin sentiment lexicons, and the process of inclusion of a prior polarity lexicon of Latin lemmas in a knowledge base of interoperable linguistic resources developed within the ERC project “LiLa: Linking Latin”. We will discuss the main challenges we face when working with ancient languages (e.g., lack of native speakers, limited amount of data, unusual textual genres for the sentiment analysis task, such as philosophical or documentary texts) and we will describe two use cases underscoring the importance of an interdisciplinary approach combining computational linguistics, semantic web and humanities practices.},
	urldate = {2021-05-21},
	author = {Sprugnoli, Rachele},
	month = mar,
	year = {2021},
	keywords = {computational linguistics, latin language, sentiment analysis},
}

While the main applications of resources and tools for sentiment analysis typically fall within the scope of fields like customer experience and social media monitoring, there is an increasing interest in extending their range to texts written in ancient and historical languages. Such interest mirrors the substantial growth of the area dedicated to building and using linguistic resources for these languages, which are essential for accessing and understanding the Classical tradition. In this talk, we will present the methodology we followed to create and evaluate a new set of Latin sentiment lexicons, and the process of inclusion of a prior polarity lexicon of Latin lemmas in a knowledge base of interoperable linguistic resources developed within the ERC project “LiLa: Linking Latin”. We will discuss the main challenges we face when working with ancient languages (e.g., lack of native speakers, limited amount of data, unusual textual genres for the sentiment analysis task, such as philosophical or documentary texts) and we will describe two use cases underscoring the importance of an interdisciplinary approach combining computational linguistics, semantic web and humanities practices.

@article{buccheri_semantic_2021,
	title = {Semantic {Analysis} and {Frequency} {Effects} of {Conceptual} {Metaphors} of {Emotions} in {Latin} : {From} a {Corpus}-{Based} {Approach} to a {Dictionary} of {Latin} {Metaphors}},
	volume = {20},
	url = {doi:10.1515/joll-2021-2002.},
	language = {English},
	number = {2},
	journal = {Journal of Latin linguistics},
	author = {Buccheri, Alessandro and De Felice, Irene and Fedriani, Chiara and Short, William M.},
	year = {2021},
	pages = {163--189},
}

@misc{korkiakangas_late_2021,
	series = {Corpora},
	title = {Late {Latin} {Charter} {Treebank}: contents and annotation},
	url = {https://researchportal.helsinki.fi/en/publications/late-latin-charter-treebank-contents-and-annotation},
	number = {16},
	author = {Korkiakangas, T.},
	year = {2021},
}

@article{czeti_structure_2021,
	title = {The structure of narrative in the story of {Baucis} and {Philemon}},
	volume = {61},
	language = {English},
	journal = {Acta Antiqua Academiae Scientiarum Hungaricae},
	author = {Czeti, István},
	year = {2021},
	pages = {243--267},
}

@incollection{tahmasebi_lexical_2021,
	series = {Language {Variation}},
	title = {Lexical semantic change for {Ancient} {Greek} and {Latin}},
	copyright = {Copyright (c) 2021 Nina Tahmasebi, Lars Borin, Adam Jatowt, Yang Xu, Simon Hengchen (Volume Editor)},
	isbn = {978-3-96110-312-6},
	url = {https://langsci-press.org/catalog/view/303/3035/2382-1},
	doi = {10.5281/zenodo.5040241},
	abstract = {Change and its precondition, variation, are inherent in languages. Over time, new words enter the lexicon, others become obsolete, and existing words acquire new senses. Associating a word with its correct meaning in its historical context is a central challenge in diachronic research. Historical corpora of classical languages, such as Ancient Greek and Latin, typically come with rich metadata, and existing models are limited by their inability to exploit contextual information beyond the document timestamp. While embedding-based methods feature among the current state of the art systems, they are lacking in their interpretative power. In contrast, Bayesian models provide explicit and interpretable representations of semantic change phenomena. In this chapter we build on GASC, a recent computational approach to semantic change based on a dynamic Bayesian mixture model. In this model, the evolution of word senses over time is based not only on distributional information of lexical nature, but also on text genres. We provide a systematic comparison of dynamic Bayesian mixture models for semantic change with state-of-the-art embedding-based models. On top of providing a full description of meaning change over time, we show that Bayesian mixture models are highly competitive approaches to detect binary semantic change in both Ancient Greek and Latin.},
	language = {en},
	number = {6},
	urldate = {2023-07-24},
	booktitle = {Language {Science} {Press}},
	publisher = {Language Science Press},
	author = {Perrone, Valerio and Hengchen, Simon and Palma, Marco and Vatri, Alessandro and Smith, Jim Q. and McGillivray, Barbara},
	editor = {Tahmasebi, Nina and Borin, Lars and Jatowt, Adam and Xu, Yang and Hengchen, Simon},
	month = feb,
	year = {2021},
	pages = {287--310},
}

Change and its precondition, variation, are inherent in languages. Over time, new words enter the lexicon, others become obsolete, and existing words acquire new senses. Associating a word with its correct meaning in its historical context is a central challenge in diachronic research. Historical corpora of classical languages, such as Ancient Greek and Latin, typically come with rich metadata, and existing models are limited by their inability to exploit contextual information beyond the document timestamp. While embedding-based methods feature among the current state of the art systems, they are lacking in their interpretative power. In contrast, Bayesian models provide explicit and interpretable representations of semantic change phenomena. In this chapter we build on GASC, a recent computational approach to semantic change based on a dynamic Bayesian mixture model. In this model, the evolution of word senses over time is based not only on distributional information of lexical nature, but also on text genres. We provide a systematic comparison of dynamic Bayesian mixture models for semantic change with state-of-the-art embedding-based models. On top of providing a full description of meaning change over time, we show that Bayesian mixture models are highly competitive approaches to detect binary semantic change in both Ancient Greek and Latin.

@article{gledic_upskills_2021,
	title = {{UPSKILLS} guidelines for {Learning} {Content} {Creation}},
	url = {https://zenodo.org/record/8302296},
	abstract = {The core of the UPSKILLS project is the production of learning content aimed at students in language- and linguistics-related fields (modern languages and cultures, translation, general linguistics, etc.) and lecturers who want to incorporate the developed content and/or add their own, into their teaching. The topics are selected in light of a comparative analysis of the current academic offer and the requirements the job market has for graduates in these areas, conducted under the UPSKILLS project. The main focus is on the knowledge and skills that are insufficiently covered in existing linguistics and language-related curricula but can open new job perspectives for students. The created learning content can be used as individual elements or as an integrated module. These guidelines are created to serve as: Reference material for UPSKILLS project partners – the partners will consult the guidelines as they create the learning content in line with the project goals. Teaching guides for those using the materials we create – those who wish to use our materials can gain insight into our approach and methodology Learning content creation guides – for those who wish to create new materials based on the model we developed under UPSKILLS},
	urldate = {2023-10-02},
	publisher = {Zenodo},
	author = {Gledić, Jelena and Assimakopoulos, Stavros and Buchberger, Iva and Budimirović, Jelena and Đukanović, Maja and Kraš, Tihana and Podboj, Martina and Soldatić, Nađa and Vella, Michela},
	month = sep,
	year = {2021},
}

The core of the UPSKILLS project is the production of learning content aimed at students in language- and linguistics-related fields (modern languages and cultures, translation, general linguistics, etc.) and lecturers who want to incorporate the developed content and/or add their own, into their teaching. The topics are selected in light of a comparative analysis of the current academic offer and the requirements the job market has for graduates in these areas, conducted under the UPSKILLS project. The main focus is on the knowledge and skills that are insufficiently covered in existing linguistics and language-related curricula but can open new job perspectives for students. The created learning content can be used as individual elements or as an integrated module. These guidelines are created to serve as: Reference material for UPSKILLS project partners – the partners will consult the guidelines as they create the learning content in line with the project goals. Teaching guides for those using the materials we create – those who wish to use our materials can gain insight into our approach and methodology Learning content creation guides – for those who wish to create new materials based on the model we developed under UPSKILLS

@article{linka_pain_2021,
	title = {Pain and the {Body} in {Corpus} {Hippocraticum}: {A} {Distributional} {Semantic} {Analysis}},
	copyright = {Copyright (c) 2021},
	issn = {2364-7957},
	shorttitle = {Pain and the {Body} in {Corpus} {Hippocraticum}},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/81212},
	doi = {10.11588/dco.2021.7.81212},
	abstract = {Die Autoren der im Corpus Hippocraticum versammelten medizinischen Abhandlungen erwähnen häufig den Schmerz, seine Eigenschaften und seinen Ursprung. Gleichzeitig liefern sie jedoch keine ausdrückliche Definition oder Theorie des Schmerzes, seiner Natur und seiner Beziehung zu anderen wichtigen Aspekten der hippokratischen Medizin. Außerdem verwenden sie mindestens vier Wortfamilien, von denen man annimmt, dass sie im Altgriechischen Schmerzen bezeichnen. Dies bringt moderne Forscher zu der Frage, wie sich diese vier Schmerzwörter semantisch unterscheiden und inwieweit sie auf einer gemeinsamen Vorstellung von Schmerz beruhen. In diesem Artikel versuchen wir, diese Fragen zu beantworten, indem wir das Korpus mit Hilfe verschiedener computergestützter Textanalysemethoden analysieren, insbesondere mit Hilfe eines Ansatzes zur distributionellen semantischen Modellierung. Unsere Ergebnisse zeigen einen engen Zusammenhang zwischen einigen dieser Schmerzwörter, Körperteilen und pathologischen Zuständen. Die Ergebnisse werden außerdem mit den Erkenntnissen verglichen, die durch traditionelles genaues Lesen der Quellen gewonnen wurden.},
	language = {en},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Linka, Vojtěch and Kaše, Vojtěch},
	month = sep,
	year = {2021},
	keywords = {DSM},
	pages = {54--71},
}

Die Autoren der im Corpus Hippocraticum versammelten medizinischen Abhandlungen erwähnen häufig den Schmerz, seine Eigenschaften und seinen Ursprung. Gleichzeitig liefern sie jedoch keine ausdrückliche Definition oder Theorie des Schmerzes, seiner Natur und seiner Beziehung zu anderen wichtigen Aspekten der hippokratischen Medizin. Außerdem verwenden sie mindestens vier Wortfamilien, von denen man annimmt, dass sie im Altgriechischen Schmerzen bezeichnen. Dies bringt moderne Forscher zu der Frage, wie sich diese vier Schmerzwörter semantisch unterscheiden und inwieweit sie auf einer gemeinsamen Vorstellung von Schmerz beruhen. In diesem Artikel versuchen wir, diese Fragen zu beantworten, indem wir das Korpus mit Hilfe verschiedener computergestützter Textanalysemethoden analysieren, insbesondere mit Hilfe eines Ansatzes zur distributionellen semantischen Modellierung. Unsere Ergebnisse zeigen einen engen Zusammenhang zwischen einigen dieser Schmerzwörter, Körperteilen und pathologischen Zuständen. Die Ergebnisse werden außerdem mit den Erkenntnissen verglichen, die durch traditionelles genaues Lesen der Quellen gewonnen wurden.

@article{nikolaev_considerations_2021,
	title = {{SOME} {CONSIDERATIONS} {ON} {THE} {ATTRIBUTION} {OF} {THE} ‘{NEW} {APULEIUS}’},
	volume = {71},
	issn = {0009-8388, 1471-6844},
	url = {https://www.cambridge.org/core/journals/classical-quarterly/article/some-considerations-on-the-attribution-of-the-new-apuleius/8F7FF4C1442452FC41CC6005E8501EB8},
	doi = {10.1017/S0009838821000987},
	abstract = {The ‘New Apuleius’ is a set of Latin summaries of Plato's works first published in 2016 by Justin Stover, who attributed it to Apuleius. The present article attempts to assess two key aspects of Stover's argument, viz. his reconstruction of the manuscript transmission of the new text and his use of computer-assisted stylometric techniques. The authors suggest that both strands of his argument are inconclusive. First, it is argued that the transposition of gatherings in the archetype of the Apuleian philosophica as envisaged by Stover is highly unrealistic. Second, replications of Stover's stylometric experiments show that their results are highly dependent on the particular algorithm settings and on the composition of the corpus. It is further shown that Stover's choice of highly specialized stylometric techniques is suboptimal, because popular generalist methods for statistical data analysis are demonstrably more successful in correctly identifying authors of Latin text fragments and do not support the case for Apuleius’ authorship of the new text. The authors conclude that there are no solid grounds to conclude that the ‘New Apuleius’ was indeed written by Apuleius.},
	language = {en},
	number = {2},
	urldate = {2023-08-26},
	journal = {The Classical Quarterly},
	publisher = {Cambridge University Press},
	author = {Nikolaev, Dmitry and Shumilin, Mikhail},
	month = dec,
	year = {2021},
	keywords = {Apuleius, Burrow's Delta, attribution, computer-assisted stylometry, transmission, ‘New Apuleius’},
	pages = {819--848},
}

The ‘New Apuleius’ is a set of Latin summaries of Plato's works first published in 2016 by Justin Stover, who attributed it to Apuleius. The present article attempts to assess two key aspects of Stover's argument, viz. his reconstruction of the manuscript transmission of the new text and his use of computer-assisted stylometric techniques. The authors suggest that both strands of his argument are inconclusive. First, it is argued that the transposition of gatherings in the archetype of the Apuleian philosophica as envisaged by Stover is highly unrealistic. Second, replications of Stover's stylometric experiments show that their results are highly dependent on the particular algorithm settings and on the composition of the corpus. It is further shown that Stover's choice of highly specialized stylometric techniques is suboptimal, because popular generalist methods for statistical data analysis are demonstrably more successful in correctly identifying authors of Latin text fragments and do not support the case for Apuleius’ authorship of the new text. The authors conclude that there are no solid grounds to conclude that the ‘New Apuleius’ was indeed written by Apuleius.

@article{nagy_carmen_2021,
	title = {Carmen et {Standard} {Error}: {Computational} {Methods} in {Stylometry} for {Classical} {Latin} {Poetry}},
	language = {en},
	author = {Nagy, Benjamin C},
	year = {2021},
}

@inproceedings{burns_profiling_2021,
	address = {Online},
	title = {Profiling of {Intertextuality} in {Latin} {Literature} {Using} {Word} {Embeddings}},
	url = {https://aclanthology.org/2021.naacl-main.389},
	doi = {10.18653/v1/2021.naacl-main.389},
	language = {en},
	urldate = {2023-08-26},
	booktitle = {Proceedings of the 2021 {Conference} of the {North} {American} {Chapter} of the {Association} for {Computational} {Linguistics}: {Human} {Language} {Technologies}},
	publisher = {Association for Computational Linguistics},
	author = {Burns, Patrick J. and Brofos, James A. and Li, Kyle and Chaudhuri, Pramit and Dexter, Joseph P.},
	year = {2021},
	pages = {4900--4907},
}

@techreport{schuller_future_2020,
	address = {Berlin},
	type = {Working {Paper}},
	title = {Future {Skills}: a {Framework} for {Data} {Literacy}},
	url = {https://hochschulforumdigitalisierung.de/sites/default/files/dateien/HFD_AP_Nr_53_Data_Literacy_Framework.pdf},
	doi = {https://doi.org/10.5281/zenodo.3946067},
	language = {en},
	number = {53},
	institution = {Hochschulforum Digitalisierung},
	author = {Schüller, Katharina},
	year = {2020},
	pages = {54},
}

@article{vaughn_participatory_2020,
	title = {Participatory {Research} {Methods} – {Choice} {Points} in the {Research} {Process}},
	volume = {1},
	url = {https://jprm.scholasticahq.com/article/13244-participatory-research-methods-choice-points-in-the-research-process},
	doi = {10.35844/001c.13244},
	abstract = {Participatory research (PR) encompasses research designs, methods, and frameworks that use systematic inquiry in direct collaboration with those affected by an issue being studied for the purpose of action or change. PR engages those who are not necessarily trained in research but belong to or represent the interests of the people who are the focus of the research. Researchers utilizing a PR approach often choose research methods and tools that can be conducted in a participatory, democratic manner that values genuine and meaningful participation in the research process. This article serves as an introduction to participatory research methods, including an overview of participatory research, terminology across disciplines, elements that make a research method participatory, and a model detailing the choice points that require decisions about which tools and methods will produce the desired level of participation at each stage of the research process. Intentional choices of participatory research methods, tools, and processes can help researchers to more meaningfully engage stakeholders and communities in research, which in turn has the potential to create relevant, meaningful research findings translated to action.},
	language = {en},
	number = {1},
	urldate = {2025-11-17},
	journal = {Journal of Participatory Research Methods},
	publisher = {Specialty Publications},
	author = {Vaughn, Lisa M. and Jacquez, Farrah},
	month = jul,
	year = {2020},
}

Participatory research (PR) encompasses research designs, methods, and frameworks that use systematic inquiry in direct collaboration with those affected by an issue being studied for the purpose of action or change. PR engages those who are not necessarily trained in research but belong to or represent the interests of the people who are the focus of the research. Researchers utilizing a PR approach often choose research methods and tools that can be conducted in a participatory, democratic manner that values genuine and meaningful participation in the research process. This article serves as an introduction to participatory research methods, including an overview of participatory research, terminology across disciplines, elements that make a research method participatory, and a model detailing the choice points that require decisions about which tools and methods will produce the desired level of participation at each stage of the research process. Intentional choices of participatory research methods, tools, and processes can help researchers to more meaningfully engage stakeholders and communities in research, which in turn has the potential to create relevant, meaningful research findings translated to action.

@inproceedings{long_what_2020,
	address = {New York, NY, USA},
	series = {{CHI} '20},
	title = {What is {AI} {Literacy}? {Competencies} and {Design} {Considerations}},
	isbn = {978-1-4503-6708-0},
	shorttitle = {What is {AI} {Literacy}?},
	url = {https://dl.acm.org/doi/pdf/10.1145/3313831.3376727},
	doi = {10.1145/3313831.3376727},
	booktitle = {Proceedings of the 2020 {CHI} {Conference} on {Human} {Factors} in {Computing} {Systems}},
	publisher = {Association for Computing Machinery},
	author = {Long, Duri and Magerko, Brian},
	year = {2020},
	keywords = {AI education, AI for K-12, AI literacy, artificial intelligence, computing education, machine learning},
	pages = {1--16},
}

@article{reddy_digital_2020,
	title = {Digital {Literacy}: {A} {Review} of {Literature}},
	volume = {11},
	issn = {1947-3451, 1947-346X},
	shorttitle = {Digital {Literacy}},
	url = {https://services.igi-global.com/resolvedoi/resolve.aspx?doi=10.4018/IJT.20200701.oa1},
	doi = {10.4018/IJT.20200701.oa1},
	abstract = {Information and communication technologies (ICT) along with the internet have fueled advancements and growth in banking, transportation, economics, and most of all in education in the 21st century. The 21st century citizens are provided with new opportunities that have been created with the advancement of ICT. Hence, individuals need a wide range of abilities, competencies, and skills to adapt to the technological era. This paper provides a literature review of the growing importance of ICT, its wide array of usage, and its influence on various facets of people’s daily lives. In addition, the emerging concept of digital literacy through ICT developments, contribution of digital literacy towards the achievement of sustainable development goals, contribution of ICT towards the development of various sectors particularly the education sector, and the work done in this area of digital literacy are summarised. The paper concludes with three new models of digital literacy: four gear model, model for flexible learning, and a model showing the impact of ICT on the learning process.},
	language = {en},
	number = {2},
	urldate = {2023-06-05},
	journal = {International Journal of Technoethics},
	author = {Reddy, Pritika and Sharma, Bibhya and Chaudhary, Kaylash},
	month = jul,
	year = {2020},
	pages = {65--94},
}

Information and communication technologies (ICT) along with the internet have fueled advancements and growth in banking, transportation, economics, and most of all in education in the 21st century. The 21st century citizens are provided with new opportunities that have been created with the advancement of ICT. Hence, individuals need a wide range of abilities, competencies, and skills to adapt to the technological era. This paper provides a literature review of the growing importance of ICT, its wide array of usage, and its influence on various facets of people’s daily lives. In addition, the emerging concept of digital literacy through ICT developments, contribution of digital literacy towards the achievement of sustainable development goals, contribution of ICT towards the development of various sectors particularly the education sector, and the work done in this area of digital literacy are summarised. The paper concludes with three new models of digital literacy: four gear model, model for flexible learning, and a model showing the impact of ICT on the learning process.

@article{fickers_update_2020,
	title = {Update für die {Hermeneutik}. {Geschichtswissenschaft} auf dem {Weg} zur digitalen {Forensik}?},
	volume = {17},
	url = {https://zeitgeschichte-digital.de/doks/frontdoor/deliver/index/docId/1765/file/ZF_1_2020_157_168_Fickers.pdf},
	doi = {10.14765/zzf.dok-1765},
	language = {de},
	number = {1},
	journal = {Zeithistorische Forschungen–Studies in Contemporary History},
	publisher = {ZZF–Centre for Contemporary History: Zeithistorische Forschungen},
	author = {Fickers, Andreas},
	year = {2020},
	pages = {157--168},
}

@misc{roberts_how_2020,
	title = {How much knowledge can you pack into the parameters of a language model?},
	url = {https://arxiv.org/pdf/2002.08910},
	publisher = {arXiv},
	author = {Roberts, Adam and Raffel, Colin and Shazeer, Noam},
	year = {2020},
}

@article{vatri_lemmatization_2020,
	title = {Lemmatization for {Ancient} {Greek}: {An} experimental assessment of the state of the art},
	volume = {20},
	issn = {1569-9846, 1566-5844},
	shorttitle = {Lemmatization for {Ancient} {Greek}},
	url = {https://brill.com/view/journals/jgl/20/2/article-p179_4.xml},
	doi = {10.1163/15699846-02002001},
	abstract = {Abstract This article presents the result of accuracy tests for currently available Ancient Greek lemmatizers and recently published lemmatized corpora. We ran a blinded experiment in which three highly proficient readers of Ancient Greek evaluated the output of the CLTK lemmatizer, of the CLTK backoff lemmatizer, and of GLEM, together with the lemmatizations offered by the Diorisis corpus and the Lemmatized Ancient Greek Texts repository. The texts chosen for this experiment are Homer, Iliad 1.1–279 and Lysias 7. The results suggest that lemmatization methods using large lexica as well as part-of-speech tagging—such as those employed by the Diorisis corpus and the CLTK backoff lemmatizer—are more reliable than methods that rely more heavily on machine learning and use smaller lexica.},
	language = {eng},
	number = {2},
	urldate = {2023-10-20},
	journal = {Journal of Greek Linguistics},
	publisher = {Brill},
	author = {Vatri, Alessandro and McGillivray, Barbara},
	month = nov,
	year = {2020},
	keywords = {Ancient Greek digital corpora, Ancient Greek morphology, digital lexicography, lemmatization, lemmatizer, usability of digital resources},
	pages = {179--196},
}

Abstract This article presents the result of accuracy tests for currently available Ancient Greek lemmatizers and recently published lemmatized corpora. We ran a blinded experiment in which three highly proficient readers of Ancient Greek evaluated the output of the CLTK lemmatizer, of the CLTK backoff lemmatizer, and of GLEM, together with the lemmatizations offered by the Diorisis corpus and the Lemmatized Ancient Greek Texts repository. The texts chosen for this experiment are Homer, Iliad 1.1–279 and Lysias 7. The results suggest that lemmatization methods using large lexica as well as part-of-speech tagging—such as those employed by the Diorisis corpus and the CLTK backoff lemmatizer—are more reliable than methods that rely more heavily on machine learning and use smaller lexica.

@article{beyer_mehr_2020,
	series = {Latein und {Griechisch} in {Nordrhein}-{Westfalen}},
	title = {Mehr {Synergie}: {Gemeinsam} zu digitalgestützten {Lehr}-{Lern}-{Settings}},
	volume = {1},
	copyright = {All rights reserved},
	issn = {2700-3531},
	url = {https://www.biejournals.de/index.php/lgnrw/article/view/5051/4652},
	doi = {10.4119/lgnrw-3983},
	number = {2},
	journal = {LGNRW},
	author = {Beyer, Andrea},
	month = dec,
	year = {2020},
	keywords = {Digitalisierung / digitalisation, didaktische Designs / educational designs, methodische Konzepte / methodological concepts},
	pages = {26--30},
}

@article{beyer_wie_2020,
	title = {Wie die {Digitalisierung} unseren {Umgang} mit den {Alten} {Sprachen} verändert hat},
	volume = {63},
	copyright = {All rights reserved},
	issn = {1432-7511},
	url = {https://journals.ub.uni-heidelberg.de/index.php/fc/article/view/78787/72743},
	doi = {10.11588/fc.2020.4.78787},
	number = {4},
	journal = {Forum Classicum},
	author = {Beyer, Andrea and Schulz, Konstantin},
	month = dec,
	year = {2020},
	keywords = {Alte Sprachen / Classical languages, Digital Humanities, Digitalisierung / digitalisation},
	pages = {208--219},
}

@article{cervantes_comprehensive_2020,
	title = {A comprehensive survey on support vector machine classification: {Applications}, challenges and trends},
	volume = {408},
	issn = {0925-2312},
	shorttitle = {A comprehensive survey on support vector machine classification},
	url = {https://www.sciencedirect.com/science/article/pii/S0925231220307153},
	doi = {10.1016/j.neucom.2019.10.118},
	abstract = {In recent years, an enormous amount of research has been carried out on support vector machines (SVMs) and their application in several fields of science. SVMs are one of the most powerful and robust classification and regression algorithms in multiple fields of application. The SVM has been playing a significant role in pattern recognition which is an extensively popular and active research area among the researchers. Research in some fields where SVMs do not perform well has spurred development of other applications such as SVM for large data sets, SVM for multi classification and SVM for unbalanced data sets. Further, SVM has been integrated with other advanced methods such as evolve algorithms, to enhance the ability of classification and optimize parameters. SVM algorithms have gained recognition in research and applications in several scientific and engineering areas. This paper provides a brief introduction of SVMs, describes many applications and summarizes challenges and trends. Furthermore, limitations of SVMs will be identified. The future of SVMs will be discussed in conjunction with further applications. The applications of SVMs will be reviewed as well, especially in the some fields.},
	urldate = {2025-10-03},
	journal = {Neurocomputing},
	author = {Cervantes, Jair and Garcia-Lamont, Farid and Rodríguez-Mazahua, Lisbeth and Lopez, Asdrubal},
	month = sep,
	year = {2020},
	keywords = {Classification, Machine learning, SVM},
	pages = {189--215},
}

In recent years, an enormous amount of research has been carried out on support vector machines (SVMs) and their application in several fields of science. SVMs are one of the most powerful and robust classification and regression algorithms in multiple fields of application. The SVM has been playing a significant role in pattern recognition which is an extensively popular and active research area among the researchers. Research in some fields where SVMs do not perform well has spurred development of other applications such as SVM for large data sets, SVM for multi classification and SVM for unbalanced data sets. Further, SVM has been integrated with other advanced methods such as evolve algorithms, to enhance the ability of classification and optimize parameters. SVM algorithms have gained recognition in research and applications in several scientific and engineering areas. This paper provides a brief introduction of SVMs, describes many applications and summarizes challenges and trends. Furthermore, limitations of SVMs will be identified. The future of SVMs will be discussed in conjunction with further applications. The applications of SVMs will be reviewed as well, especially in the some fields.

@article{freund_forschendes_2020,
	title = {Forschendes {Lernen} im {Praxissemester} unter den {Bedingungen} kleiner {Fächer}: {Ein} {Praxiskonzept} für die {Begleitung} von {Studienprojekten} im {Praxissemester} am {Beispiel} des {Faches} {Latein}},
	volume = {2},
	copyright = {Copyright (c) 2020 Stefan Freund, Leoni Janssen},
	issn = {2629-5598},
	shorttitle = {Forschendes {Lernen} im {Praxissemester} unter den {Bedingungen} kleiner {Fächer}},
	url = {https://www.dimawe.de/index.php/dimawe/article/view/3900},
	doi = {10.4119/dimawe-3900},
	abstract = {Die Betreuung von Studienprojekten im Rahmen des Praxissemesters, in denen Studierende in einer Verzahnung von Theorie und Praxis forschendes Lernen betreiben sollen, stellt kleine Fächer in der universitären Leh­rerbildung vor besondere Herausforderungen. Der vorliegende Beitrag zeigt am Beispiel des Faches Latein ein Praxismodell für den Ablauf der Praxissemesterstudienprojektbetreuung, der forschendes Lernen durch begleitete Formulierung der Forschungsfrage und begleitende Reflexionsimpulse unterstützt. Hinzu kommen als Sonderfälle die zeitliche Distanz zwischen Vorbereitungsveranstaltung und Praxissemester, die personelle Diskontinuität bei der Betreuung und die Einbindung von Promovierenden.},
	language = {de},
	number = {2},
	urldate = {2023-07-16},
	journal = {Die Materialwerkstatt. Zeitschrift für Konzepte und Arbeitsmaterialien für Lehrer*innenbildung und Unterricht.},
	author = {Freund, Stefan and Janssen, Leoni},
	month = nov,
	year = {2020},
	note = {Number: 2},
	keywords = {Kleine Fächer, Lehrerbildung, Praxisphasen, Praxissemester, Studienprojekt, forschendes Lernen},
	pages = {66--74},
}

Die Betreuung von Studienprojekten im Rahmen des Praxissemesters, in denen Studierende in einer Verzahnung von Theorie und Praxis forschendes Lernen betreiben sollen, stellt kleine Fächer in der universitären Leh­rerbildung vor besondere Herausforderungen. Der vorliegende Beitrag zeigt am Beispiel des Faches Latein ein Praxismodell für den Ablauf der Praxissemesterstudienprojektbetreuung, der forschendes Lernen durch begleitete Formulierung der Forschungsfrage und begleitende Reflexionsimpulse unterstützt. Hinzu kommen als Sonderfälle die zeitliche Distanz zwischen Vorbereitungsveranstaltung und Praxissemester, die personelle Diskontinuität bei der Betreuung und die Einbindung von Promovierenden.

@inproceedings{sprugnoli_overview_2020,
	address = {Marseille, France},
	title = {Overview of the {EvaLatin} 2020 {Evaluation} {Campaign}},
	isbn = {979-10-95546-53-5},
	url = {https://aclanthology.org/2020.lt4hala-1.16},
	abstract = {This paper describes the first edition of EvaLatin, a campaign totally devoted to the evaluation of NLP tools for Latin. The two shared tasks proposed in EvaLatin 2020, i. e. Lemmatization and Part-of-Speech tagging, are aimed at fostering research in the field of language technologies for Classical languages. The shared dataset consists of texts taken from the Perseus Digital Library, processed with UDPipe models and then manually corrected by Latin experts. The training set includes only prose texts by Classical authors. The test set, alongside with prose texts by the same authors represented in the training set, also includes data relative to poetry and to the Medieval period. This also allows us to propose the Cross-genre and Cross-time subtasks for each task, in order to evaluate the portability of NLP tools for Latin across different genres and time periods. The results obtained by the participants for each task and subtask are presented and discussed.},
	language = {English},
	urldate = {2024-08-31},
	booktitle = {Proceedings of {LT4HALA} 2020 - 1st {Workshop} on {Language} {Technologies} for {Historical} and {Ancient} {Languages}},
	publisher = {European Language Resources Association (ELRA)},
	author = {Sprugnoli, Rachele and Passarotti, Marco and Cecchini, Flavio Massimiliano and Pellegrini, Matteo},
	editor = {Sprugnoli, Rachele and Passarotti, Marco},
	month = may,
	year = {2020},
	pages = {105--110},
}

This paper describes the first edition of EvaLatin, a campaign totally devoted to the evaluation of NLP tools for Latin. The two shared tasks proposed in EvaLatin 2020, i. e. Lemmatization and Part-of-Speech tagging, are aimed at fostering research in the field of language technologies for Classical languages. The shared dataset consists of texts taken from the Perseus Digital Library, processed with UDPipe models and then manually corrected by Latin experts. The training set includes only prose texts by Classical authors. The test set, alongside with prose texts by the same authors represented in the training set, also includes data relative to poetry and to the Medieval period. This also allows us to propose the Cross-genre and Cross-time subtasks for each task, in order to evaluate the portability of NLP tools for Latin across different genres and time periods. The results obtained by the participants for each task and subtask are presented and discussed.

@article{uglanova_order_2020,
	title = {The {Order} of {Things}. {A} {Study} on {Topic} {Modelling} of {Literary} {Texts}.},
	url = {https://ceur-ws.org/Vol-2723/long7.pdf},
	number = {18-20},
	journal = {CHR},
	author = {Uglanova, Inna and Gius, Evelyn and Karsdorp, F and McGillivray, B and Nerghes, A and Wevers, M},
	year = {2020},
	pages = {2020},
}

@inproceedings{palladino_ner_2020,
	title = {{NER} on {Ancient} {Greek} with minimal annotation},
	url = {https://works.hcommons.org/records/wv2mz-m9p64},
	doi = {10.17613/j7jt-b052},
	abstract = {This paper presents the results in the adaptation of a new workflow of Named Entity Recognition and classification applied to Ancient Greek. We used a model of data extraction and pattern discovery based on machine learning algorithms which is easily customizable for different languages. This allowed the creation of a dataset of automatically classified place-names and ethnonyms starting from a small manually annotated list. We worked on the assumption that premodern textual sources display a recognized systematicity in their linguistic encoding of space, which provides a test-case for automatic context-based methods. The idea is that we should be able to train the machine to recognize an entity from recurring elements in the context, without providing a large training dataset in advance.},
	language = {en},
	urldate = {2025-06-17},
	booktitle = {https://dh2020.adho.org/},
	publisher = {DH2020},
	author = {Palladino, Chiara and Karimi, Farimah and Mathiak, Brigitte},
	year = {2020},
}

This paper presents the results in the adaptation of a new workflow of Named Entity Recognition and classification applied to Ancient Greek. We used a model of data extraction and pattern discovery based on machine learning algorithms which is easily customizable for different languages. This allowed the creation of a dataset of automatically classified place-names and ethnonyms starting from a small manually annotated list. We worked on the assumption that premodern textual sources display a recognized systematicity in their linguistic encoding of space, which provides a test-case for automatic context-based methods. The idea is that we should be able to train the machine to recognize an entity from recurring elements in the context, without providing a large training dataset in advance.

@inproceedings{boros_comparison_2020,
	title = {A comparison of sequential and combined approaches for named entity recognition in a corpus of handwritten medieval charters},
	isbn = {1-7281-9966-2},
	url = {https://storage.teklia.com/teklia-public-website/documents/ICFHR2020_NER_Comparison_final_updated_Q0Qt4Te.pdf},
	publisher = {IEEE},
	author = {Boroş, Emanuela and Romero, Verónica and Maarand, Martin and Zenklová, Kateřina and Křečková, Jitka and Vidal, Enrique and Stutzmann, Dominique and Kermorvant, Christopher},
	year = {2020},
	pages = {79--84},
}

@article{de_carvalho-filho_twelve_2020,
	title = {Twelve tips for implementing a community of practice for faculty development},
	volume = {42},
	issn = {0142-159X},
	url = {https://doi.org/10.1080/0142159X.2018.1552782},
	doi = {10.1080/0142159X.2018.1552782},
	abstract = {Teaching and learning practices often fail to incorporate new concepts in the ever-evolving field of medical education. Although medical education research provides new insights into curricular development, learners’ engagement, assessment methods, professional development, interprofessional education, and so forth, faculty members often struggle to modernize their teaching practices. Communities of practice (CoP) for faculty development offer an effective and sustainable approach for knowledge management and implementation of best practices. A successful CoP creates and shares knowledge in the context of a specific practice toward the development of expertise. CoPs’ collaborative nature, based on the co-creation of practical solutions to daily problems, aligns well with the goals of applying best practices in health professions education and training new faculty members. In our article, we share 12 tips for implementing a community of practice for faculty development. The tips were based on a comprehensive literature review and the authors’ experiences.},
	number = {2},
	urldate = {2025-02-02},
	journal = {Medical Teacher},
	publisher = {Taylor \& Francis},
	author = {de Carvalho-Filho, Marco Antonio and Tio, René A. and Steinert, Yvonne},
	month = feb,
	year = {2020},
	note = {\_eprint: https://doi.org/10.1080/0142159X.2018.1552782},
	pages = {143--149},
}

Teaching and learning practices often fail to incorporate new concepts in the ever-evolving field of medical education. Although medical education research provides new insights into curricular development, learners’ engagement, assessment methods, professional development, interprofessional education, and so forth, faculty members often struggle to modernize their teaching practices. Communities of practice (CoP) for faculty development offer an effective and sustainable approach for knowledge management and implementation of best practices. A successful CoP creates and shares knowledge in the context of a specific practice toward the development of expertise. CoPs’ collaborative nature, based on the co-creation of practical solutions to daily problems, aligns well with the goals of applying best practices in health professions education and training new faculty members. In our article, we share 12 tips for implementing a community of practice for faculty development. The tips were based on a comprehensive literature review and the authors’ experiences.

@inproceedings{foka_semantically_2020,
	address = {Seattle, Washington, USA},
	title = {Semantically geo-annotating an ancient {Greek} "travel guide" {Itineraries}, {Chronotopes}, {Networks}, and {Linked} {Data}},
	isbn = {978-1-4503-8163-5},
	url = {https://dl.acm.org/doi/10.1145/3423337.3429433},
	doi = {10.1145/3423337.3429433},
	abstract = {Pausanias's second-century CE Periegesis Hellados presents a ten-volume grand tour of the Greek mainland. After the post-enlightenment rediscovery of ancient Greek literature, his Description of Greece proved highly influential as a guidebook to Greece's antiquities, directing travellers and archaeologists alike to uncovering and interpreting major sites, notably at Athens, Corinth and Olympia. Recent studies focusing on his Description as a narrative, however, have drawn attention to the textual construction of space, and the different ways in which space and place are conceptualised and related to each other. This paper outlines the initial work of the Digital Periegesis project, which is using semantic geo-annotation to capture and analyse the forms of space within and the spatial form of this narrative. In particular, it discusses the challenges and affordances of using geo-parsing, spatio-temporal analysis, network analysis, and Linked Open Data (LOD) for rethinking the geographies of a non-modern literary text as based more on topological connections than topographic proximity.},
	language = {en},
	urldate = {2025-01-26},
	booktitle = {Proceedings of the 4th {ACM} {SIGSPATIAL} {Workshop} on {Geospatial} {Humanities}},
	publisher = {ACM},
	author = {Foka, Anna and Barker, Elton and Konstantinidou, Kyriaki and Mostofian, Nasrin and Demiroglu, O. Cenk and Kiesling, Brady and Talatas, Linda},
	year = {2020},
	pages = {1--9},
}

Pausanias's second-century CE Periegesis Hellados presents a ten-volume grand tour of the Greek mainland. After the post-enlightenment rediscovery of ancient Greek literature, his Description of Greece proved highly influential as a guidebook to Greece's antiquities, directing travellers and archaeologists alike to uncovering and interpreting major sites, notably at Athens, Corinth and Olympia. Recent studies focusing on his Description as a narrative, however, have drawn attention to the textual construction of space, and the different ways in which space and place are conceptualised and related to each other. This paper outlines the initial work of the Digital Periegesis project, which is using semantic geo-annotation to capture and analyse the forms of space within and the spatial form of this narrative. In particular, it discusses the challenges and affordances of using geo-parsing, spatio-temporal analysis, network analysis, and Linked Open Data (LOD) for rethinking the geographies of a non-modern literary text as based more on topological connections than topographic proximity.

@misc{berra_aurelberrastopwords_2020,
	title = {aurelberra/stopwords v2.3.0},
	copyright = {Open Access},
	url = {https://zenodo.org/record/1165205},
	doi = {10.5281/ZENODO.1165205},
	abstract = {This repository contains Ancient Greek and Latin stopwords for textual analysis.},
	urldate = {2024-05-17},
	publisher = {[object Object]},
	author = {Berra, Aurélien},
	month = may,
	year = {2020},
}

This repository contains Ancient Greek and Latin stopwords for textual analysis.

@techreport{stoeckel_voting_2020,
	title = {Voting for {POS} {Tagging} of {Latin} {Texts} : {Using} the {Flair} of {FLAIR} to {Better} {Ensemble} {Classifiers} by {Example} of {Latin}},
	url = {https://aclanthology.org/2020.lt4hala-1.21.pdf},
	number = {Proceedings of the LREC 2020 1st Workshop on Language Technologies for Historical and Ancient Languages (LT4HALA 2020)},
	author = {Stoeckel, Manuel and Henlein, Alexander and Hemati, Wahed and Mehler, Alexander},
	year = {2020},
	pages = {130--135},
}

@inproceedings{straka_udpipe_2020,
	title = {{UDPipe} at {EvaLatin} 2020: {Contextualized} embeddings and treebank embeddings},
	url = {https://arxiv.org/pdf/2006.03687.pdf},
	booktitle = {Proceedings of {LT4HALA} 2020-1st {Workshop} on {Language} {Technologies} for {Historical} and {Ancient} {Languages}},
	author = {Straka, Milan and Straková, Jana},
	year = {2020},
	pages = {124--129},
}

@article{gorman_author_2020,
	title = {Author {Identification} of {Short} {Texts} {Using} {Dependency} {Treebanks} without {Vocabulary}},
	volume = {35},
	doi = {doi:10.1093/llc/fqz070},
	number = {4},
	journal = {Digital scholarship in the humanities},
	author = {Gorman, Robert J.},
	year = {2020},
	pages = {812--825},
}

@inproceedings{yeruva_interpretation_2020,
	address = {Online},
	title = {Interpretation of {Sentiment} {Analysis} in {Aeschylus}'s {Greek} {Tragedy}},
	url = {https://www.aclweb.org/anthology/2020.latechclfl-1.17},
	abstract = {Recent advancements in NLP and machine learning have created unique challenges and opportunities for digital humanities research. In particular, there are ample opportunities for NLP and machine learning researchers to analyze data from literary texts and to broaden our understanding of human sentiment in classical Greek tragedy. In this paper, we will explore the challenges and benefits from the human and machine collaboration for sentiment analysis in Greek tragedy and address some open questions related to the collaborative annotation for the sentiments in literary texts. We focus primarily on (i) an analysis of the challenges in sentiment analysis tasks for humans and machines, and (ii) whether consistent annotation results are generated from the multiple human annotators and multiple machine annotators. For human annotators, we have used a survey-based approach with about 60 college students. We have selected three popular sentiment analysis tools for machine annotators, including VADER, CoreNLP's sentiment annotator, and TextBlob. We have conducted a qualitative and quantitative evaluation and confirmed our observations on sentiments in Greek tragedy.},
	urldate = {2021-05-21},
	booktitle = {Proceedings of the {The} 4th {Joint} {SIGHUM} {Workshop} on {Computational} {Linguistics} for {Cultural} {Heritage}, {Social} {Sciences}, {Humanities} and {Literature}},
	publisher = {International Committee on Computational Linguistics},
	author = {Yeruva, Vijaya Kumari and ChandraShekar, Mayanka and Lee, Yugyung and Rydberg-Cox, Jeff and Blanton, Virginia and Oyler, Nathan A},
	month = dec,
	year = {2020},
	pages = {138--146},
}

Recent advancements in NLP and machine learning have created unique challenges and opportunities for digital humanities research. In particular, there are ample opportunities for NLP and machine learning researchers to analyze data from literary texts and to broaden our understanding of human sentiment in classical Greek tragedy. In this paper, we will explore the challenges and benefits from the human and machine collaboration for sentiment analysis in Greek tragedy and address some open questions related to the collaborative annotation for the sentiments in literary texts. We focus primarily on (i) an analysis of the challenges in sentiment analysis tasks for humans and machines, and (ii) whether consistent annotation results are generated from the multiple human annotators and multiple machine annotators. For human annotators, we have used a survey-based approach with about 60 college students. We have selected three popular sentiment analysis tools for machine annotators, including VADER, CoreNLP's sentiment annotator, and TextBlob. We have conducted a qualitative and quantitative evaluation and confirmed our observations on sentiments in Greek tragedy.

@inproceedings{yeruva_interpretation_2020,
	title = {Interpretation of {Sentiment} {Analysis} with {Human}-in-the-{Loop}},
	isbn = {1-72816-251-3},
	publisher = {IEEE},
	author = {Yeruva, Vijaya Kumari and Chandrashekar, Mayanka and Lee, Yugyung and Rydberg-Cox, Jeff and Blanton, Virginia and Oyler, Nathan A},
	year = {2020},
	pages = {3099--3108},
}

@inproceedings{sprugnoli_odi_2020,
	title = {Odi et {Amo}. {Creating}, {Evaluating} and {Extending} {Sentiment} {Lexicons} for {Latin}.},
	url = {https://aclanthology.org/2020.lrec-1.376.pdf},
	author = {Sprugnoli, Rachele and Passarotti, Marco and Corbetta, Daniela and Peverelli, Andrea},
	year = {2020},
	pages = {3078--3086},
}

@inproceedings{hellwig_treebank_2020,
	address = {Marseille, France},
	title = {The {Treebank} of {Vedic} {Sanskrit}},
	url = {https://aclanthology.org/2020.lrec-1.632/},
	booktitle = {Proceedings of the {Twelfth} {Language} {Resources} and {Evaluation} {Conference}},
	publisher = {European Language Resources Association},
	author = {Hellwig, Oliver and Scarlata, Salvatore and Widmer, Paul},
	year = {2020},
	pages = {5137--5146},
}

@article{papantoniou_nlp_2020,
	title = {{NLP} for the {Greek} {Language}: {A} {Brief} {Survey}.},
	doi = {https://doi.org/10.1145/3411408.3411410},
	journal = {11th Hellenic Conference on Artificial Intelligence (SETN 2020)},
	author = {Papantoniou, Katerina and Tzitzikas, Yannis},
	year = {2020},
}

@techreport{pellegrini_using_2020,
	title = {Using {LatInfLexi} for an {Entropy}-{Based} {Assessment} of {Predictability} in {Latin} {Inflection}},
	number = {Proceedings of the LREC 2020 1st Workshop on Language Technologies for Historical and Ancient Languages (LT4HALA 2020)},
	author = {Pellegrini, Matteo},
	year = {2020},
	pages = {37--46},
}

@article{vayansky_review_2020,
	title = {A review of topic modeling methods},
	volume = {94},
	url = {https://www.researchgate.net/profile/Sathish_Kumar50/publication/342288300_A_review_of_topic_modeling_methods/links/5ef14381299bf1faac6f22f9/A-review-of-topic-modeling-methods.pdf},
	doi = {10.1016/j.is.2020.101582},
	abstract = {Topic modeling is a popular analytical tool for evaluating data. Numerous methods of topic modeling have been developed which consider many kinds of relationships and restrictions within datasets; however, these methods are not frequently employed. Instead many researchers gravitate to Latent Dirichlet Analysis, which although flexible and adaptive, is not always suited for modeling more complex data relationships. We present different topic modeling approaches capable of dealing with correlation between topics, the changes of topics over time, as well as the ability to handle short texts such as encountered in social media or sparse text data. We also briefly review the algorithms which are used to optimize and infer parameters in topic modeling, which is essential to producing meaningful results regardless of method. We believe this review will encourage more diversity when performing topic modeling and help determine what topic modeling method best suits the user needs.},
	journal = {Information Systems},
	author = {Vayansky, Ike and Kumar, Sathish},
	month = jun,
	year = {2020},
	pages = {1--32},
}

Topic modeling is a popular analytical tool for evaluating data. Numerous methods of topic modeling have been developed which consider many kinds of relationships and restrictions within datasets; however, these methods are not frequently employed. Instead many researchers gravitate to Latent Dirichlet Analysis, which although flexible and adaptive, is not always suited for modeling more complex data relationships. We present different topic modeling approaches capable of dealing with correlation between topics, the changes of topics over time, as well as the ability to handle short texts such as encountered in social media or sparse text data. We also briefly review the algorithms which are used to optimize and infer parameters in topic modeling, which is essential to producing meaningful results regardless of method. We believe this review will encourage more diversity when performing topic modeling and help determine what topic modeling method best suits the user needs.

@book{florio-hansen_digitalisierung_2020,
	title = {Digitalisierung, {Künstliche} {Intelligenz} und {Robotik}: {Eine} {Einführung} für {Schule} und {Unterricht}},
	isbn = {978-3-8252-5429-2},
	shorttitle = {Digitalisierung, {Künstliche} {Intelligenz} und {Robotik}},
	url = {https://books.google.it/books?hl=en&lr=&id=HHUGEAAAQBAJ},
	language = {de},
	publisher = {UTB},
	author = {Florio-Hansen, Inez De},
	month = nov,
	year = {2020},
	note = {Google-Books-ID: HHUGEAAAQBAJ},
}

@book{richards_fundamentals_2020,
	title = {Fundamentals of {Software} {Architecture}: {An} {Engineering} {Approach}},
	isbn = {978-1-4920-4342-3},
	shorttitle = {Fundamentals of {Software} {Architecture}},
	url = {https://books.google.de/books?hl=de&lr=&id=xa7MDwAAQBAJ},
	abstract = {Salary surveys worldwide regularly place software architect in the top 10 best jobs, yet no real guide exists to help developers become architects. Until now. This book provides the first comprehensive overview of software architecture’s many aspects. Aspiring and existing architects alike will examine architectural characteristics, architectural patterns, component determination, diagramming and presenting architecture, evolutionary architecture, and many other topics.Mark Richards and Neal Ford—hands-on practitioners who have taught software architecture classes professionally for years—focus on architecture principles that apply across all technology stacks. You’ll explore software architecture in a modern light, taking into account all the innovations of the past decade.This book examines:Architecture patterns: The technical basis for many architectural decisionsComponents: Identification, coupling, cohesion, partitioning, and granularitySoft skills: Effective team management, meetings, negotiation, presentations, and moreModernity: Engineering practices and operational approaches that have changed radically in the past few yearsArchitecture as an engineering discipline: Repeatable results, metrics, and concrete valuations that add rigor to software architecture},
	language = {en},
	publisher = {"O'Reilly Media, Inc."},
	author = {Richards, Mark and Ford, Neal},
	month = jan,
	year = {2020},
	note = {Google-Books-ID: xa7MDwAAQBAJ},
	keywords = {Computers / Software Development \& Engineering / General, Computers / Software Development \& Engineering / Systems Analysis \& Design, Computers / Software Development \& Engineering / Tools, Computers / Systems Architecture / Distributed Systems \& Computing},
}

Salary surveys worldwide regularly place software architect in the top 10 best jobs, yet no real guide exists to help developers become architects. Until now. This book provides the first comprehensive overview of software architecture’s many aspects. Aspiring and existing architects alike will examine architectural characteristics, architectural patterns, component determination, diagramming and presenting architecture, evolutionary architecture, and many other topics.Mark Richards and Neal Ford—hands-on practitioners who have taught software architecture classes professionally for years—focus on architecture principles that apply across all technology stacks. You’ll explore software architecture in a modern light, taking into account all the innovations of the past decade.This book examines:Architecture patterns: The technical basis for many architectural decisionsComponents: Identification, coupling, cohesion, partitioning, and granularitySoft skills: Effective team management, meetings, negotiation, presentations, and moreModernity: Engineering practices and operational approaches that have changed radically in the past few yearsArchitecture as an engineering discipline: Repeatable results, metrics, and concrete valuations that add rigor to software architecture

@article{diemke_alkibiades_2020,
	title = {Alkibiades, {Pyrrhos} und {Alexander}: {Eine} {Untersuchung} zu {Emotionen} und {Gewalt} in den {Viten} {Plutarchs} unter {Verwendung} digitaler {Methoden}},
	copyright = {Copyright (c) 2020 Digital Classics Online},
	issn = {2364-7957},
	shorttitle = {Alkibiades, {Pyrrhos} und {Alexander}},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/77663},
	doi = {10.11588/dco.2020.2.77663},
	abstract = {Forscher haben die Bedeutung von Emotionen in Plutarchs Biographien weitgehend ignoriert, obwohl Emotionen für die Entstehung von Gewalt eine entscheidende Rolle spielen. Mit Hilfe von ERIS, einem Hamburger Informationssystem zur Darstellung griechischer und römischer Gewalt, werden Gewaltdarstellungen, die auf ein emotionales Motiv zurückgehen, in den Biographien von Alkibiades, Pyrrhos und Alexander untersucht. Durch eine Visualisierung lassen sich Muster und Beziehungen zwischen den Objekten und Merkmalen schneller erkennen. Die Ergebnisse zeigen, wie digitale Werkzeuge dazu beitragen können, neue Beziehungen zwischen Opfer, Täter, Waffe, Gewaltmethode und Motiv aufzudecken. Darüber hinaus zeigen die Ergebnisse, wie stark der wachsende Machteinfluss und das Fehlen von Paideia das Gewaltverhalten und die fehlende Selbstkontrolle der Protagonisten forcieren können. Die Untersuchung soll das Erkenntnispotenzial und den Mehrwert, der aus der Anwendung von digitaler und hermeneutischer Analyse resultiert, aufzeigen.},
	language = {de},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Diemke, Justine},
	month = dec,
	year = {2020},
	keywords = {Informationssystem},
	pages = {57--74},
}

Forscher haben die Bedeutung von Emotionen in Plutarchs Biographien weitgehend ignoriert, obwohl Emotionen für die Entstehung von Gewalt eine entscheidende Rolle spielen. Mit Hilfe von ERIS, einem Hamburger Informationssystem zur Darstellung griechischer und römischer Gewalt, werden Gewaltdarstellungen, die auf ein emotionales Motiv zurückgehen, in den Biographien von Alkibiades, Pyrrhos und Alexander untersucht. Durch eine Visualisierung lassen sich Muster und Beziehungen zwischen den Objekten und Merkmalen schneller erkennen. Die Ergebnisse zeigen, wie digitale Werkzeuge dazu beitragen können, neue Beziehungen zwischen Opfer, Täter, Waffe, Gewaltmethode und Motiv aufzudecken. Darüber hinaus zeigen die Ergebnisse, wie stark der wachsende Machteinfluss und das Fehlen von Paideia das Gewaltverhalten und die fehlende Selbstkontrolle der Protagonisten forcieren können. Die Untersuchung soll das Erkenntnispotenzial und den Mehrwert, der aus der Anwendung von digitaler und hermeneutischer Analyse resultiert, aufzeigen.

@inproceedings{nicolosi_clarin-it_2020,
	title = {{CLARIN}-{IT} and the {Definition} of a {Digital} {Critical} {Edition} for {Ancient} {Greek} {Poetry}},
	url = {https://ep.liu.se/en/conference-article.aspx?series=ecp&issue=172&Article_No=11},
	doi = {10.3384/ecp2020172011},
	abstract = {Ancient Greek studies, and Classics in general, is a perfect field of investigation in Digital Humanities. Indeed, DH approaches could become a means of building models for complex realities, analyzing them with computational methods and sharing the results with a broader public. Ancient texts have a complex tradition, which includes many witnesses (texts that handed down other texts) and different typologies of supports (papyri, manuscripts, and epigraphs). These texts are the basis of all European Literatures and it is crucial to spread their knowledge, in a reliable and easy way. Our project on ancient Greek fragmentary poetry (DEA - Digital Edition of Archilochus: New models and tools for authoring, editing and indexing an ancient Greek fragmentary author), growing out of the existing experience, tries to define a TEI-based digital critical edition combined with NLP techniques and semantic web technologies. Our goal is to provide a complete and reliable tool for scholars, suitable for critical studies in Classics, and a user-friendly environment also for non-specialist users. The project represents one of the attempts within the context of CLARIN-IT to contribute to the wider impact of CLARIN on the specific Italian community interested in Digital Classics. It is intended to improve services in fostering new knowledge in SSH digital research and sustaining the existing one.},
	language = {en},
	urldate = {2023-08-26},
	author = {Nicolosi, Anika and Monachini, Monica and Nova, Beatrice},
	month = jul,
	year = {2020},
	pages = {85--93},
}

Ancient Greek studies, and Classics in general, is a perfect field of investigation in Digital Humanities. Indeed, DH approaches could become a means of building models for complex realities, analyzing them with computational methods and sharing the results with a broader public. Ancient texts have a complex tradition, which includes many witnesses (texts that handed down other texts) and different typologies of supports (papyri, manuscripts, and epigraphs). These texts are the basis of all European Literatures and it is crucial to spread their knowledge, in a reliable and easy way. Our project on ancient Greek fragmentary poetry (DEA - Digital Edition of Archilochus: New models and tools for authoring, editing and indexing an ancient Greek fragmentary author), growing out of the existing experience, tries to define a TEI-based digital critical edition combined with NLP techniques and semantic web technologies. Our goal is to provide a complete and reliable tool for scholars, suitable for critical studies in Classics, and a user-friendly environment also for non-specialist users. The project represents one of the attempts within the context of CLARIN-IT to contribute to the wider impact of CLARIN on the specific Italian community interested in Digital Classics. It is intended to improve services in fostering new knowledge in SSH digital research and sustaining the existing one.

@article{ribary_corpus_2020,
	title = {A {Corpus} {Approach} to {Roman} {Law} {Based} on {Justinian}’s {Digest}},
	volume = {7},
	copyright = {http://creativecommons.org/licenses/by/3.0/},
	issn = {2227-9709},
	url = {https://www.mdpi.com/2227-9709/7/4/44},
	doi = {10.3390/informatics7040044},
	abstract = {Traditional philological methods in Roman legal scholarship such as close reading and strict juristic reasoning have analysed law in extraordinary detail. Such methods, however, have paid less attention to the empirical characteristics of legal texts and occasionally projected an abstract framework onto the sources. The paper presents a series of computer-assisted methods to open new frontiers of inquiry. Using a Python coding environment, we have built a relational database of the Latin text of the Digest, a historical sourcebook of Roman law compiled under the order of Emperor Justinian in 533 CE. Subsequently, we investigated the structure of Roman law by automatically clustering the sections of the Digest according to their linguistic profile. Finally, we explored the characteristics of Roman legal language according to the principles and methods of computational distributional semantics. Our research has discovered an empirical structure of Roman law which arises from the sources themselves and complements the dominant scholarly assumption that Roman law rests on abstract structures. By building and comparing Latin word embeddings models, we were also able to detect a semantic split in words with general and legal sense. These investigations point to a practical focus in Roman law which is consistent with the view that ancient law schools were more interested in training lawyers for practice rather than in philosophical neatness.},
	language = {en},
	number = {4},
	urldate = {2023-08-26},
	journal = {Informatics},
	publisher = {Multidisciplinary Digital Publishing Institute},
	author = {Ribary, Marton and McGillivray, Barbara},
	month = dec,
	year = {2020},
	note = {Number: 4},
	keywords = {Digest, Latin, LatinISE, Python, Roman law, clustering, computational linguistics, corpus linguistics, distributional semantics, word embeddings},
	pages = {44},
}

Traditional philological methods in Roman legal scholarship such as close reading and strict juristic reasoning have analysed law in extraordinary detail. Such methods, however, have paid less attention to the empirical characteristics of legal texts and occasionally projected an abstract framework onto the sources. The paper presents a series of computer-assisted methods to open new frontiers of inquiry. Using a Python coding environment, we have built a relational database of the Latin text of the Digest, a historical sourcebook of Roman law compiled under the order of Emperor Justinian in 533 CE. Subsequently, we investigated the structure of Roman law by automatically clustering the sections of the Digest according to their linguistic profile. Finally, we explored the characteristics of Roman legal language according to the principles and methods of computational distributional semantics. Our research has discovered an empirical structure of Roman law which arises from the sources themselves and complements the dominant scholarly assumption that Roman law rests on abstract structures. By building and comparing Latin word embeddings models, we were also able to detect a semantic split in words with general and legal sense. These investigations point to a practical focus in Roman law which is consistent with the view that ancient law schools were more interested in training lawyers for practice rather than in philosophical neatness.

@article{burns_ensemble_2020,
	title = {Ensemble lemmatization with the {Classical} {Language} {Toolkit}},
	volume = {58},
	copyright = {Copyright (c) 2020 Studi e Saggi Linguistici},
	issn = {2281-9142},
	url = {https://studiesaggilinguistici.it/ssl/article/view/273},
	doi = {10.4454/ssl.v58i1.273},
	abstract = {Because of the less-resourced nature of historical languages, non-standard solutions are often required for natural language processing tasks. This article introduces one such solution for historical-language lemmatization, that is the Ensemble lemmatizer for the Classical Language Toolkit, an open-source Python package that supports NLP research for historical languages. Ensemble lemmatization is the most recent development at CLTK in the repurposing and refactoring of an existing method designed for one task, specifically the backoff method as used for part-of-speech tagging, for use in a different task, namely lemmatization. This article argues for the benefits of ensemble lemmatization, specifically, flexible tool construction and the use of all available information to reach tagging decisions, and presents two use cases.},
	language = {en},
	number = {1},
	urldate = {2023-08-26},
	journal = {Studi e Saggi Linguistici},
	author = {Burns, Patrick J.},
	month = sep,
	year = {2020},
	note = {Number: 1},
	pages = {157--176},
}

Because of the less-resourced nature of historical languages, non-standard solutions are often required for natural language processing tasks. This article introduces one such solution for historical-language lemmatization, that is the Ensemble lemmatizer for the Classical Language Toolkit, an open-source Python package that supports NLP research for historical languages. Ensemble lemmatization is the most recent development at CLTK in the repurposing and refactoring of an existing method designed for one task, specifically the backoff method as used for part-of-speech tagging, for use in a different task, namely lemmatization. This article argues for the benefits of ensemble lemmatization, specifically, flexible tool construction and the use of all available information to reach tagging decisions, and presents two use cases.

@article{sprugnoli_building_2020,
	title = {Building and {Comparing} {Lemma} {Embeddings} for {Latin}. {Classical} {Latin} versus {Thomas} {Aquinas}},
	volume = {6},
	copyright = {https://creativecommons.org/licenses/by-nc-nd/4.0/},
	issn = {2499-4553},
	url = {https://journals.openedition.org/ijcol/624},
	doi = {10.4000/ijcol.624},
	abstract = {This paper presents a new set of lemma embeddings for the Latin language. Embeddings are trained on a manually annotated corpus of texts belonging to the Classical era: different models, architectures and dimensions are tested and evaluated using a novel benchmark for the synonym selection task. In addition, we release vectors pre-trained on the “Opera Maiora” by Thomas Aquinas, thus providing a resource to analyze Latin in a diachronic perspective. The embeddings built upon the two training corpora are compared to each other to support diachronic lexical studies. The words showing the highest usage change between the two corpora are reported and a selection of them is discussed.},
	language = {en},
	number = {1},
	urldate = {2023-08-26},
	journal = {IJCoL. Italian Journal of Computational Linguistics},
	publisher = {Accademia University Press},
	author = {Sprugnoli, Rachele and Moretti, Giovanni and Passarotti, Marco},
	month = jun,
	year = {2020},
	note = {Number: 1},
	pages = {29--45},
}

This paper presents a new set of lemma embeddings for the Latin language. Embeddings are trained on a manually annotated corpus of texts belonging to the Classical era: different models, architectures and dimensions are tested and evaluated using a novel benchmark for the synonym selection task. In addition, we release vectors pre-trained on the “Opera Maiora” by Thomas Aquinas, thus providing a resource to analyze Latin in a diachronic perspective. The embeddings built upon the two training corpora are compared to each other to support diachronic lexical studies. The words showing the highest usage change between the two corpora are reported and a selection of them is discussed.

@misc{bamman_latin_2020,
	title = {Latin {BERT}: {A} {Contextual} {Language} {Model} for {Classical} {Philology}},
	shorttitle = {Latin {BERT}},
	url = {http://arxiv.org/abs/2009.10053},
	abstract = {We present Latin BERT, a contextual language model for the Latin language, trained on 642.7 million words from a variety of sources spanning the Classical era to the 21st century. In a series of case studies, we illustrate the affordances of this language-specific model both for work in natural language processing for Latin and in using computational methods for traditional scholarship: we show that Latin BERT achieves a new state of the art for part-of-speech tagging on all three Universal Dependency datasets for Latin and can be used for predicting missing text (including critical emendations); we create a new dataset for assessing word sense disambiguation for Latin and demonstrate that Latin BERT outperforms static word embeddings; and we show that it can be used for semanticallyinformed search by querying contextual nearest neighbors. We publicly release trained models to help drive future work in this space.},
	language = {en},
	urldate = {2023-08-26},
	publisher = {arXiv},
	author = {Bamman, David and Burns, Patrick J.},
	month = sep,
	year = {2020},
	note = {arXiv:2009.10053 [cs]},
	keywords = {Computer Science - Computation and Language},
}

We present Latin BERT, a contextual language model for the Latin language, trained on 642.7 million words from a variety of sources spanning the Classical era to the 21st century. In a series of case studies, we illustrate the affordances of this language-specific model both for work in natural language processing for Latin and in using computational methods for traditional scholarship: we show that Latin BERT achieves a new state of the art for part-of-speech tagging on all three Universal Dependency datasets for Latin and can be used for predicting missing text (including critical emendations); we create a new dataset for assessing word sense disambiguation for Latin and demonstrate that Latin BERT outperforms static word embeddings; and we show that it can be used for semanticallyinformed search by querying contextual nearest neighbors. We publicly release trained models to help drive future work in this space.

@book{greenham_close_2019,
	address = {Boca Raton, FL},
	edition = {First edition},
	series = {Seminar {Studies}},
	title = {Close reading: {The} basics},
	isbn = {978-0-203-70997-9},
	url = {https://www.taylorfrancis.com/books/9780203709979},
	abstract = {chapter Introduction – chapter 1 Seven pleasures of reading – chapter 2 The semantic context – chapter 3 The syntactic and thematic contexts – chapter 4 Getting to the crux of things: Hamlet, a case study – chapter 5 The iterative context, Part 1: character and plot – chapter 6 The iterative context, Part 2: sounds and rhythms – chapter 7 The generic context – chapter 8 The adversarial context.},
	publisher = {Routledge},
	author = {Greenham, David},
	year = {2019},
	note = {Backup Publisher: Taylor and Francis},
}

chapter Introduction – chapter 1 Seven pleasures of reading – chapter 2 The semantic context – chapter 3 The syntactic and thematic contexts – chapter 4 Getting to the crux of things: Hamlet, a case study – chapter 5 The iterative context, Part 1: character and plot – chapter 6 The iterative context, Part 2: sounds and rhythms – chapter 7 The generic context – chapter 8 The adversarial context.

@article{schuller_future_2019,
	title = {Future {Skills}: {Ein} {Framework} für {Data} {Literacy}},
	volume = {13},
	issn = {1863-8155, 1863-8163},
	url = {http://link.springer.com/10.1007/s11943-019-00261-9},
	doi = {10.1007/s11943-019-00261-9},
	language = {de},
	number = {3-4},
	urldate = {2023-04-19},
	journal = {AStA Wirtschafts- und Sozialstatistisches Archiv},
	author = {Schüller, Katharina and Busch, Paulina and Hindinger, Carina},
	month = nov,
	year = {2019},
	keywords = {A20, D80, D83, Datenethik, Datenkompetenz, Datenkultur, Kompetenzrahmen, M14, Statistical Literacy},
	pages = {297--317},
}

@inproceedings{sprugnoli_vir_2019,
	address = {Bari},
	title = {\textit{{Vir}} is to \textit{{Moderatus}} as \textit{{Mulier}} is to \textit{{Intemperans}}. {Lemma} {Embeddings} for {Latin}},
	url = {https://ceur-ws.org/Vol-2481/paper69.pdf},
	doi = {urn:nbn:de:0074-2481-7},
	booktitle = {Proceedings of the {Sixth} {Italian} {Conference} on {Computational} {Linguistics}},
	author = {Sprugnoli, Rachele and Passarotti, Marco and Moretti, Giovanni},
	year = {2019},
}

@article{oliver_faith_2019,
	title = {Faith leaders developing digital literacies: {Demands} and resources across career stages according to theological educators},
	volume = {11},
	issn = {2167-8715},
	url = {https://digitalcommons.uri.edu/cgi/viewcontent.cgi?article=1409&context=jmle},
	doi = {10.23860/JMLE-2019-11-2-7},
	number = {2},
	journal = {Journal of Media Literacy Education},
	author = {Oliver, Kyle M and Williams-Duncan, Stacy},
	year = {2019},
	pages = {122--145},
}

@book{huber_vom_2019,
	address = {Wiesbaden},
	title = {Vom forschungsnahen zum forschenden {Lernen} an {Hochschulen}: {Wege} der {Bildung} durch {Wissenschaft}},
	copyright = {http://www.springer.com/tdm},
	isbn = {978-3-658-24948-9 978-3-658-24949-6},
	shorttitle = {Vom forschungsnahen zum forschenden {Lernen} an {Hochschulen}},
	url = {http://link.springer.com/10.1007/978-3-658-24949-6},
	doi = {10.1007/978-3-658-24949-6},
	language = {de},
	urldate = {2024-09-01},
	publisher = {Springer Fachmedien},
	author = {Huber, Ludwig and Reinmann, Gabi},
	year = {2019},
	keywords = {Erziehungswissenschaft, Hochschulbildung, Hochschuldidaktik, Schlüsselkompetenzen, Scholarship of Teaching and Learning, Wissenschaft, Wissenschaftliches Prüfen, learning and instruction},
}

@incollection{berti_ecomparatio_2019,
	title = {{eComparatio} – a {Software} {Tool} for {Automatic} {Text} {Comparison}},
	copyright = {http://creativecommons.org/licenses/by-nc-nd/4.0},
	isbn = {978-3-11-059957-2},
	url = {https://www.degruyterbrill.com/document/doi/10.1515/9783110599572-013/html},
	doi = {10.1515/9783110599572-013},
	abstract = {The following paper gives a short description of the software-tool eComparatio that was originally intended as a tool for the comparison of different text editions. An example of its original purposes will be given, the larger part of the paper consists of a detailed description of the actual comparison process in detail. In a final section, some differences to similar text comparison tools for plain text will be given.},
	language = {en},
	urldate = {2025-10-25},
	booktitle = {Digital {Classical} {Philology}},
	publisher = {De Gruyter},
	author = {Bräckel, Oliver and Kahl, Hannes and Meins, Friedrich and Schubert, Charlotte},
	editor = {Berti, Monica},
	month = aug,
	year = {2019},
	pages = {221--238},
}

The following paper gives a short description of the software-tool eComparatio that was originally intended as a tool for the comparison of different text editions. An example of its original purposes will be given, the larger part of the paper consists of a detailed description of the actual comparison process in detail. In a final section, some differences to similar text comparison tools for plain text will be given.

@book{dobler_prufungsregime_2019,
	address = {Wiesbaden},
	title = {Prüfungsregime und {Prüfungskulturen}: {Soziologische} {Beobachtungen} zur internen {Organisation} von {Hochschule}},
	isbn = {978-3-658-25289-2},
	shorttitle = {Prüfungsregime und {Prüfungskulturen}},
	url = {http://link.springer.com/10.1007/978-3-658-25290-8},
	doi = {10.1007/978-3-658-25290-8},
	language = {de},
	urldate = {2023-05-17},
	publisher = {Springer Fachmedien},
	author = {Döbler, Joachim},
	year = {2019},
	keywords = {Bologna-Prozess, Constructive Alignment, Hochschulverwaltung, Interne Organisation von Hochschule, Lernforschung, Notengebung, Prüfungskulturen, Prüfungspraxis, Prüfungsregime, Prüfungswesen, Qualifikationsrahmen},
}

@book{bakker_design_2019,
	address = {New York},
	title = {Design {Research} in {Education}: {A} {Practical} {Guide} for {Early} {Career} {Researchers}},
	isbn = {978-1-351-32941-5},
	url = {https://books.google.de/books?id=6jhjDwAAQBAJ},
	publisher = {Routledge},
	author = {Bakker, Arthur},
	year = {2019},
	lccn = {2020691622},
}

@article{trust_six_2019,
	title = {Six {Key} {Elements} {Identified} in an {Active} and {Thriving} {Blended} {Community} of {Practice}},
	volume = {63},
	issn = {1559-7075},
	url = {https://doi.org/10.1007/s11528-018-0265-x},
	doi = {10.1007/s11528-018-0265-x},
	abstract = {Communities of practice offer situated, social and distributed learning experiences that support the professional growth of teachers. However, while some communities of practice are successful, others flounder. This manuscript reports the findings of a qualitative study that examined teachers’ experiences in the Discovery Educator Network, a blended community of practice that positively impacted teacher growth, in order to identify the elements that shaped teacher participation and learning. Data was collected through in-depth interviews with 26 members of the Discovery Educator Network. Data analysis resulted in the identification of six key elements that were critical to the success of the community of practice: leadership roles, personalized learning, guiding principles, organizational support, social learning and purpose.},
	language = {en},
	number = {2},
	urldate = {2023-07-16},
	journal = {TechTrends},
	author = {Trust, Torrey and Horrocks, Brian},
	month = mar,
	year = {2019},
	keywords = {Community of practice, Connected learning, Discovery education, Professional development, Social media, Teacher learning},
	pages = {108--115},
}

Communities of practice offer situated, social and distributed learning experiences that support the professional growth of teachers. However, while some communities of practice are successful, others flounder. This manuscript reports the findings of a qualitative study that examined teachers’ experiences in the Discovery Educator Network, a blended community of practice that positively impacted teacher growth, in order to identify the elements that shaped teacher participation and learning. Data was collected through in-depth interviews with 26 members of the Discovery Educator Network. Data analysis resulted in the identification of six key elements that were critical to the success of the community of practice: leadership roles, personalized learning, guiding principles, organizational support, social learning and purpose.

@article{kim_survey_2019,
	series = {Zeitschrift für digitale {Geisteswissenschaften}},
	title = {A {Survey} on {Sentiment} and {Emotion} {Analysis} for                                 {Computational} {Literary} {Studies}},
	copyright = {CC BY-SA 4.0},
	url = {http://dx.doi.org/10.17175/2019_008_v2},
	language = {en},
	author = {Kim, Evgeny and Klinger, Roman},
	month = dec,
	year = {2019},
}

@article{kaplan_siri_2019,
	title = {Siri, {Siri}, in my hand: {Who}’s the fairest in the land? {On} the interpretations, illustrations, and implications of artificial intelligence},
	volume = {62},
	issn = {0007-6813},
	shorttitle = {Siri, {Siri}, in my hand},
	url = {https://www.sciencedirect.com/science/article/pii/S0007681318301393},
	doi = {10.1016/j.bushor.2018.08.004},
	abstract = {Artificial intelligence (AI)—defined as a system’s ability to correctly interpret external data, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaptation—is a topic in nearly every boardroom and at many dinner tables. Yet, despite this prominence, AI is still a surprisingly fuzzy concept and a lot of questions surrounding it are still open. In this article, we analyze how AI is different from related concepts, such as the Internet of Things and big data, and suggest that AI is not one monolithic term but instead needs to be seen in a more nuanced way. This can either be achieved by looking at AI through the lens of evolutionary stages (artificial narrow intelligence, artificial general intelligence, and artificial super intelligence) or by focusing on different types of AI systems (analytical AI, human-inspired AI, and humanized AI). Based on this classification, we show the potential and risk of AI using a series of case studies regarding universities, corporations, and governments. Finally, we present a framework that helps organizations think about the internal and external implications of AI, which we label the Three C Model of Confidence, Change, and Control.},
	number = {1},
	urldate = {2025-03-15},
	journal = {Business Horizons},
	author = {Kaplan, Andreas and Haenlein, Michael},
	month = jan,
	year = {2019},
	keywords = {Artificial intelligence, Big data, Deep learning, Expert systems, Internet of Things, Machine learning},
	pages = {15--25},
}

Artificial intelligence (AI)—defined as a system’s ability to correctly interpret external data, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaptation—is a topic in nearly every boardroom and at many dinner tables. Yet, despite this prominence, AI is still a surprisingly fuzzy concept and a lot of questions surrounding it are still open. In this article, we analyze how AI is different from related concepts, such as the Internet of Things and big data, and suggest that AI is not one monolithic term but instead needs to be seen in a more nuanced way. This can either be achieved by looking at AI through the lens of evolutionary stages (artificial narrow intelligence, artificial general intelligence, and artificial super intelligence) or by focusing on different types of AI systems (analytical AI, human-inspired AI, and humanized AI). Based on this classification, we show the potential and risk of AI using a series of case studies regarding universities, corporations, and governments. Finally, we present a framework that helps organizations think about the internal and external implications of AI, which we label the Three C Model of Confidence, Change, and Control.

@article{wang_evaluating_2019,
	title = {Evaluating word embedding models: methods and experimental results},
	url = {https://www.cambridge.org/core/services/aop-cambridge-core/content/view/EDF43F837150B94E71DBB36B28B85E79/S204877031900012Xa.pdf/div-class-title-evaluating-word-embedding-models-methods-and-experimental-results-div.pdf},
	doi = {doi:10.1017/ATSIP.2019.12},
	abstract = {Extensive evaluation on a large number of word embedding models for language processing applications is conducted in this
work. First, we introduce popular word embedding models and discuss desired properties of word models and evaluation methods
(or evaluators). Then, we categorize evaluators into intrinsic and extrinsic two types. Intrinsic evaluators test the quality of a
representation independent of specific natural language processing tasks while extrinsic evaluators use word embeddings as input
features to a downstream task and measure changes in performance metrics specific to that task. We report experimental results
of intrinsic and extrinsic evaluators on six word embedding models. It is shown that different evaluators focus on different aspects
of word models, and some are more correlated with natural language processing tasks. Finally, we adopt correlation analysis to
study performance consistency of extrinsic and intrinsic evaluators.},
	author = {Wang, Bin and Wang, Angela and Chen, Fenxiao and Wang, Yuncheng and Kuo, C.-C. Jay},
	year = {2019},
	keywords = {Word Embeddings},
}

Extensive evaluation on a large number of word embedding models for language processing applications is conducted in this work. First, we introduce popular word embedding models and discuss desired properties of word models and evaluation methods (or evaluators). Then, we categorize evaluators into intrinsic and extrinsic two types. Intrinsic evaluators test the quality of a representation independent of specific natural language processing tasks while extrinsic evaluators use word embeddings as input features to a downstream task and measure changes in performance metrics specific to that task. We report experimental results of intrinsic and extrinsic evaluators on six word embedding models. It is shown that different evaluators focus on different aspects of word models, and some are more correlated with natural language processing tasks. Finally, we adopt correlation analysis to study performance consistency of extrinsic and intrinsic evaluators.

@incollection{burns_building_2019,
	address = {Berlin \& Boston},
	title = {Building a {Text} {Analysis} {Pipeline} for {Classical} {Languages}},
	url = {https://doi.org/10.1515/9783110599572-010},
	booktitle = {Digital {Classical} {Philology}: {Ancient} {Greek} and {Latin} in the {Digital} {Revolution}},
	publisher = {De Gruyter},
	author = {Burns, Patrick J},
	editor = {Berti, Monica},
	year = {2019},
	pages = {159--176},
}

@article{min_modeling_2019,
	title = {Modeling narrative structure and dynamics with networks, sentiment analysis, and topic modeling},
	volume = {14},
	url = {https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0226025},
	number = {12},
	journal = {PLoS ONE},
	author = {Min, Semi and Park, Juyong},
	year = {2019},
}

@inproceedings{mitchell_model_2019,
	title = {Model {Cards} for {Model} {Reporting}},
	url = {http://arxiv.org/abs/1810.03993},
	doi = {10.1145/3287560.3287596},
	abstract = {Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment. In order to clarify the intended use cases of machine learning models and minimize their usage in contexts for which they are not well suited, we recommend that released models be accompanied by documentation detailing their performance characteristics. In this paper, we propose a framework that we call model cards, to encourage such transparent model reporting. Model cards are short documents accompanying trained machine learning models that provide benchmarked evaluation in a variety of conditions, such as across different cultural, demographic, or phenotypic groups (e.g., race, geographic location, sex, Fitzpatrick skin type) and intersectional groups (e.g., age and race, or sex and Fitzpatrick skin type) that are relevant to the intended application domains. Model cards also disclose the context in which models are intended to be used, details of the performance evaluation procedures, and other relevant information. While we focus primarily on human-centered machine learning models in the application fields of computer vision and natural language processing, this framework can be used to document any trained machine learning model. To solidify the concept, we provide cards for two supervised models: One trained to detect smiling faces in images, and one trained to detect toxic comments in text. We propose model cards as a step towards the responsible democratization of machine learning and related AI technology, increasing transparency into how well AI technology works. We hope this work encourages those releasing trained machine learning models to accompany model releases with similar detailed evaluation numbers and other relevant documentation.},
	urldate = {2024-06-01},
	booktitle = {Proceedings of the {Conference} on {Fairness}, {Accountability}, and {Transparency}},
	author = {Mitchell, Margaret and Wu, Simone and Zaldivar, Andrew and Barnes, Parker and Vasserman, Lucy and Hutchinson, Ben and Spitzer, Elena and Raji, Inioluwa Deborah and Gebru, Timnit},
	month = jan,
	year = {2019},
	note = {arXiv:1810.03993 [cs]},
	keywords = {Computer Science - Artificial Intelligence, Computer Science - Machine Learning},
	pages = {220--229},
}

Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment. In order to clarify the intended use cases of machine learning models and minimize their usage in contexts for which they are not well suited, we recommend that released models be accompanied by documentation detailing their performance characteristics. In this paper, we propose a framework that we call model cards, to encourage such transparent model reporting. Model cards are short documents accompanying trained machine learning models that provide benchmarked evaluation in a variety of conditions, such as across different cultural, demographic, or phenotypic groups (e.g., race, geographic location, sex, Fitzpatrick skin type) and intersectional groups (e.g., age and race, or sex and Fitzpatrick skin type) that are relevant to the intended application domains. Model cards also disclose the context in which models are intended to be used, details of the performance evaluation procedures, and other relevant information. While we focus primarily on human-centered machine learning models in the application fields of computer vision and natural language processing, this framework can be used to document any trained machine learning model. To solidify the concept, we provide cards for two supervised models: One trained to detect smiling faces in images, and one trained to detect toxic comments in text. We propose model cards as a step towards the responsible democratization of machine learning and related AI technology, increasing transparency into how well AI technology works. We hope this work encourages those releasing trained machine learning models to accompany model releases with similar detailed evaluation numbers and other relevant documentation.

@article{rodda_vector_2019,
	title = {Vector space models of {Ancient} {Greek} word meaning, and a case study on {Homer}},
	volume = {60},
	issn = {1248-9433},
	url = {https://aclanthology.org/2019.tal-3.4.pdf},
	number = {3},
	journal = {Traitement Automatique des Langues},
	publisher = {Lavoisier},
	author = {Rodda, M and Probert, Philomen and McGillivray, Barbara},
	year = {2019},
}

@inproceedings{devlin_bert_2019,
	address = {Minneapolis, Minnesota},
	title = {{BERT}: {Pre}-training of {Deep} {Bidirectional} {Transformers} for {Language} {Understanding}},
	shorttitle = {{BERT}},
	url = {https://www.aclweb.org/anthology/N19-1423},
	doi = {10.18653/v1/N19-1423},
	abstract = {We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models (Peters et al., 2018a; Radford et al., 2018), BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5 (7.7 point absolute improvement), MultiNLI accuracy to 86.7\% (4.6\% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).},
	urldate = {2020-11-25},
	booktitle = {Proceedings of the 2019 {Conference} of the {North} {American} {Chapter} of the {Association} for {Computational} {Linguistics}: {Human} {Language} {Technologies}, {Volume} 1 ({Long} and {Short} {Papers})},
	publisher = {Association for Computational Linguistics},
	author = {Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
	month = jun,
	year = {2019},
	pages = {4171--4186},
}

We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models (Peters et al., 2018a; Radford et al., 2018), BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5 (7.7 point absolute improvement), MultiNLI accuracy to 86.7% (4.6% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).

@inproceedings{bolt_stylometry_2019,
	address = {Hong Kong, China},
	title = {A {Stylometry} {Toolkit} for {Latin} {Literature}},
	booktitle = {Proceedings of the 2019 {Conference} on {Empirical} {Methods} in {Natural} {Language} {Processing} and the 9th {International} {Joint} {Conference} on {Natural} {Language} {Processing} ({EMNLP}-{IJCNLP}): {System} {Demonstrations}},
	publisher = {Association for Computational Linguistics},
	author = {Bolt, Thomas J. and Flynt, Jeffrey H. and Chaudhuri, Pramit and Dexter, Joseph P},
	year = {2019},
	pages = {205--210},
}

@book{schubert_platon_2019,
	title = {Platon digital: {Tradition} und {Rezeption}},
	isbn = {978-3-947450-07-7},
	shorttitle = {Platon digital},
	url = {https://nbn-resolving.org/urn:nbn:de:101:1-2019070314102240510161},
	abstract = {Platon ist nach Homer der antike Autor mit der reichhaltigsten Rezeption vom Altertum über das Mittelalter bis in die Neuzeit. Gleichwohl und gerade aus diesem Grund ist diese bisher allenfalls bruchstückhaft aufgearbeitet worden. Die Autoren versuchen, diesem alten Ziel geisteswissenschaftlicher Forschung auf neuen Wegen näherzukommen, indem sie eine informationswissenschaftliche Perspektive auf Platon und seine Rezeption anwenden. Dazu sind innovative Methoden der Paraphrasensuche entwickelt worden, um diese auch als Methode altertumswissenschaftlich und kulturwissenschaftlich interessierter Forschung zu etablieren.},
	language = {German},
	urldate = {2020-02-17},
	author = {Schubert, Charlotte and Molitor, Paul and Ritter, Jörg and Scharloth, Joachim and Sier, Kurt},
	year = {2019},
	keywords = {reference},
}

Platon ist nach Homer der antike Autor mit der reichhaltigsten Rezeption vom Altertum über das Mittelalter bis in die Neuzeit. Gleichwohl und gerade aus diesem Grund ist diese bisher allenfalls bruchstückhaft aufgearbeitet worden. Die Autoren versuchen, diesem alten Ziel geisteswissenschaftlicher Forschung auf neuen Wegen näherzukommen, indem sie eine informationswissenschaftliche Perspektive auf Platon und seine Rezeption anwenden. Dazu sind innovative Methoden der Paraphrasensuche entwickelt worden, um diese auch als Methode altertumswissenschaftlich und kulturwissenschaftlich interessierter Forschung zu etablieren.

@book{berti_digital_2019,
	address = {Berlin},
	series = {Age of {Access}? {Grundfragen} der {Informationsgesellschaft}},
	title = {Digital classical philology: {Ancient} {Greek} and {Latin} in the digital revolution},
	volume = {10},
	isbn = {978-3-11-059678-6},
	abstract = {Thanks to the digital revolution, even a traditional discipline like philology has been enjoying a renaissance within academia and beyond. Decades of work have been producing groundbreaking results, raising new research questions and creating innovative educational resources. This book describes the rapidly developing state of the art of digital philology with a focus on Ancient Greek and Latin, the classical languages of Western culture. Contributions cover a wide range of topics about the accessibility and analysis of Greek and Latin sources. The discussion is organized in five sections concerning open data of Greek and Latin texts; catalogs and citations of authors and works; data entry, collection and analysis for classical philology; critical editions and annotations of sources; and finally linguistic annotations and lexical databases. As a whole, the volume provides a comprehensive outline of an emergent research field for a new generation of scholars and students, explaining what is reachable and analyzable that was not before in terms of technology and accessibility.},
	publisher = {Walter de Gruyter \& Co},
	author = {Berti, Monica},
	year = {2019},
	keywords = {Humanités digitales, Numérisation, Philologie classique, reference},
}

Thanks to the digital revolution, even a traditional discipline like philology has been enjoying a renaissance within academia and beyond. Decades of work have been producing groundbreaking results, raising new research questions and creating innovative educational resources. This book describes the rapidly developing state of the art of digital philology with a focus on Ancient Greek and Latin, the classical languages of Western culture. Contributions cover a wide range of topics about the accessibility and analysis of Greek and Latin sources. The discussion is organized in five sections concerning open data of Greek and Latin texts; catalogs and citations of authors and works; data entry, collection and analysis for classical philology; critical editions and annotations of sources; and finally linguistic annotations and lexical databases. As a whole, the volume provides a comprehensive outline of an emergent research field for a new generation of scholars and students, explaining what is reachable and analyzable that was not before in terms of technology and accessibility.

@article{vainio_reconsidering_2019,
	title = {Reconsidering {Authorship} in the {Ciceronian} {Corpus} through {Computational} {Authorship} {Attribution}},
	volume = {3},
	url = {https://ojs.unito.it/index.php/COL/article/view/3518/3182},
	language = {en},
	number = {1},
	journal = {Ciceroniana online},
	author = {Vainio, Raija and Välimäki, Reima and Vesanto, Aleksi and Hella, Anni and Kaartinen, Marjo and Immonen, Teemu},
	year = {2019},
	pages = {15--48},
}

@techreport{keersmaekers_creating_2019,
	title = {Creating, {Enriching} and {Valorizing} {Treebanks} of {Ancient} {Greek}.},
	url = {https://syntaxfest.github.io/syntaxfest19/proceedings/papers/paper_68.pdf},
	institution = {Association for Computational Linguistics (ACL)},
	author = {Keersmaekers, Alek and Mercelis, Wouter and Swaelens, Colin and Toon, Van Hal},
	year = {2019},
	pages = {109--117},
}

@misc{beyer_teaching_2019,
	address = {Berlin},
	type = {Talk},
	title = {Teaching {Digital} {Literacy} – {Interpretieren} in einer computergestützten {Lehr}-/{Lernumgebung}},
	url = {https://doi.org/10.5281/zenodo.3674815},
	doi = {10.5281/zenodo.3674815},
	language = {Deutsch},
	author = {Beyer, Andrea and Reichetanz, Paul},
	month = mar,
	year = {2019},
	keywords = {reference},
}

@inproceedings{franzini_nunc_2019,
	title = {Nunc {Est} {Aestimandum}: {Towards} an {Evaluation} of the {Latin} {WordNet}},
	url = {https://www.researchgate.net/profile/Greta-Franzini-2/publication/336799230_Nunc_Est_Aestimandum_Towards_an_Evaluation_of_the_Latin_WordNet/links/5db2be42299bf111d4c83184/Nunc-Est-Aestimandum-Towards-an-Evaluation-of-the-Latin-WordNet.pdf},
	author = {Franzini, Greta and Peverelli, Andrea and Ruffolo, Paolo and Passarotti, Marco and Sanna, Helena and Signoroni, Edoardo and Ventura, Viviana and Zampedri, Federica},
	year = {2019},
}

@incollection{pockelmann_word_2019,
	title = {Word {Mover}’s {Distance} angewendet auf die {Paraphrasenextraktion} im {Altgriechischen}},
	url = {https://books.ub.uni-heidelberg.de/propylaeum/reader/download/451/451-30-84795-1-10-20190507.pdf},
	booktitle = {Platon {Digital}. {Tradition} und {Rezeption}},
	publisher = {Propylaeum Heidelberg},
	author = {Pöckelmann, Marcus and Ritter, Jörg and Molitor, Paul},
	editor = {Schubert, Charlotte and Molitor, Paul and Ritter, Jörg and Sier, Kurt and Scharloth, Joachim},
	year = {2019},
	pages = {45--60},
}

@article{schubert_visualisierung_2019,
	title = {Visualisierung von {Textdaten}: {Die} {Falle} der {Metadaten} am {Beispiel} von {Iamblichs} {Protreptikos}},
	copyright = {Copyright (c) 2019 Digital Classics Online},
	issn = {2364-7957},
	shorttitle = {Visualisierung von {Textdaten}},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/59356},
	doi = {10.11588/dco.2019.1.59356},
	abstract = {„Digital Humanities analysieren nicht nur Bilder, sondern produzieren auch neue Bilder“[1] – diese alltägliche Feststellung beleuchtet einen Prozeß, dessen Verlauf einerseits noch ganz offen ist, da diese neuen Repräsentationsmöglichkeiten epistemisch keineswegs erfaßt sind, geschweige denn, daß die Entwicklung in diesem Bereich zu stabilen Praktiken geführt hätte. Andererseits zeigt sich ein unhinterfragter Siegeslauf, der auch schon zu einem neuen Feld wie dem der Visualization Literacy geführt hat.
Im vorliegenden Beitrag wird die Visualisierung anhand von Metadaten untersucht. Gerade die Metadaten sind heute im Kontext der großen Datenmengen, die als ‚Big Data‘ anfallen von größter Bedeutung. Die für die Analyse von Big Data notwendigen Aggregationen von Daten sind ohne Metadaten nicht effizient und leistungsstark durchzuführen. Über die normale Datenerfassung hinaus legen Metadaten Muster offen, die sonst nicht sichtbar wären. Dies wiederum wird über ‚Verbildlichung’ als einer heute gängigen Repräsentationsform ermöglicht: Gerade auch für Texte gilt, daß sie in praktischen Anwendungen auf der Grundlage ihrer Metadaten durch Visualisierung zu Bildern werden, die quantitativ ausgewertet können und so wiederum auch in den Forschungsdiskurs eingehen. Von diesen gängigen Vorgehen ausgehend, stellt sich die Frage, ob diese gegenwärtige Praxis wissenschaftlichen Ansprüchen genügt oder ob sich derzeit nicht vielmehr – in einer Zeit, in der Daten als das Öl oder Gold des 21. Jahrhunderts betrachtet werden – eine Art Goldgräberstimmung und ein entsprechend unkritisches Verhalten etablieren.
 
[1] Kwastek 2015; vgl. Kath et al. 2015.},
	language = {de},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Schubert, Charlotte},
	month = jun,
	year = {2019},
	keywords = {Protreptikos},
	pages = {4--21},
}

„Digital Humanities analysieren nicht nur Bilder, sondern produzieren auch neue Bilder“[1] – diese alltägliche Feststellung beleuchtet einen Prozeß, dessen Verlauf einerseits noch ganz offen ist, da diese neuen Repräsentationsmöglichkeiten epistemisch keineswegs erfaßt sind, geschweige denn, daß die Entwicklung in diesem Bereich zu stabilen Praktiken geführt hätte. Andererseits zeigt sich ein unhinterfragter Siegeslauf, der auch schon zu einem neuen Feld wie dem der Visualization Literacy geführt hat. Im vorliegenden Beitrag wird die Visualisierung anhand von Metadaten untersucht. Gerade die Metadaten sind heute im Kontext der großen Datenmengen, die als ‚Big Data‘ anfallen von größter Bedeutung. Die für die Analyse von Big Data notwendigen Aggregationen von Daten sind ohne Metadaten nicht effizient und leistungsstark durchzuführen. Über die normale Datenerfassung hinaus legen Metadaten Muster offen, die sonst nicht sichtbar wären. Dies wiederum wird über ‚Verbildlichung’ als einer heute gängigen Repräsentationsform ermöglicht: Gerade auch für Texte gilt, daß sie in praktischen Anwendungen auf der Grundlage ihrer Metadaten durch Visualisierung zu Bildern werden, die quantitativ ausgewertet können und so wiederum auch in den Forschungsdiskurs eingehen. Von diesen gängigen Vorgehen ausgehend, stellt sich die Frage, ob diese gegenwärtige Praxis wissenschaftlichen Ansprüchen genügt oder ob sich derzeit nicht vielmehr – in einer Zeit, in der Daten als das Öl oder Gold des 21. Jahrhunderts betrachtet werden – eine Art Goldgräberstimmung und ein entsprechend unkritisches Verhalten etablieren.   [1] Kwastek 2015; vgl. Kath et al. 2015.

@article{riess_violence_2019,
	title = {Violence and the {Sea}: {A} {Digital} {Analysis} of {Maritime} {Acts} of {Violence} {Committed} by {Alcibiades} as {Described} by {Thucydides}, {Xenophon}, and {Plutarch}},
	copyright = {Copyright (c) 2020},
	issn = {2364-7957},
	shorttitle = {Violence and the {Sea}},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/72018},
	doi = {10.11588/dco.2019.2.72018},
	abstract = {Beim Vergleich der Gewaltmuster, die von Alkibiades in den Werken von Thukydides, Plutarch und Xenophon ausgeübt wurden, treten signifikante Unterschiede im Fokus der drei Autoren zutage, die durch "Eris. Das Hamburger Informationssystem über die Darstellung griechischer und römischer Gewalt" visuell dargestellt werden können. Die Tatsache, dass die Interpretation der graphischen Befunde – also der Kategorien zweiter Ordnung – dennoch zu sehr plausiblen Ergebnissen führt, zeigt, dass dieser Proof of Concept erfolgreich war. Die plausiblen Ergebnisse legen auch nahe, dass wir durch die Verwendung von Eris bisher unentdeckte Gewaltmuster bei der Untersuchung der großen Daten antiker Texte finden werden. Die Grafiken werden somit als Inspirationsquellen dienen, die neue Fragen aufwerfen, die aufgrund der großen Datenmengen noch nicht in unser Denken eingedrungen sind.},
	language = {en},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Riess, Werner},
	year = {2019},
	keywords = {Visualisierung},
	pages = {4--27},
}

Beim Vergleich der Gewaltmuster, die von Alkibiades in den Werken von Thukydides, Plutarch und Xenophon ausgeübt wurden, treten signifikante Unterschiede im Fokus der drei Autoren zutage, die durch "Eris. Das Hamburger Informationssystem über die Darstellung griechischer und römischer Gewalt" visuell dargestellt werden können. Die Tatsache, dass die Interpretation der graphischen Befunde – also der Kategorien zweiter Ordnung – dennoch zu sehr plausiblen Ergebnissen führt, zeigt, dass dieser Proof of Concept erfolgreich war. Die plausiblen Ergebnisse legen auch nahe, dass wir durch die Verwendung von Eris bisher unentdeckte Gewaltmuster bei der Untersuchung der großen Daten antiker Texte finden werden. Die Grafiken werden somit als Inspirationsquellen dienen, die neue Fragen aufwerfen, die aufgrund der großen Datenmengen noch nicht in unser Denken eingedrungen sind.

@article{chaudhuri_small_2019,
	title = {A small set of stylometric features differentiates {Latin} prose and verse},
	volume = {34},
	issn = {2055-7671},
	url = {https://doi.org/10.1093/llc/fqy070},
	doi = {10.1093/llc/fqy070},
	abstract = {Identifying the stylistic signatures characteristic of different genres is of central importance to literary theory and criticism. In this article we report a large-scale computational analysis of Latin prose and verse using a combination of quantitative stylistics and supervised machine learning. We train a set of classifiers to differentiate prose and poetry with high accuracy (\>97\%) based on a set of twenty-six text-based, primarily syntactic features and rank the relative importance of these features to identify a low-dimensional set still sufficient to achieve excellent classifier performance. This analysis demonstrates that Latin prose and verse can be classified effectively using just three top features. From examination of the highly ranked features, we observe that measures of the hypotactic style favored in Latin prose (i.e. subordinating constructions in complex sentences, such as relative clauses) are especially useful for classification.},
	number = {4},
	urldate = {2023-08-26},
	journal = {Digital Scholarship in the Humanities},
	author = {Chaudhuri, Pramit and Dasgupta, Tathagata and Dexter, Joseph P and Iyer, Krithika},
	month = dec,
	year = {2019},
	pages = {716--729},
}

Identifying the stylistic signatures characteristic of different genres is of central importance to literary theory and criticism. In this article we report a large-scale computational analysis of Latin prose and verse using a combination of quantitative stylistics and supervised machine learning. We train a set of classifiers to differentiate prose and poetry with high accuracy (>97%) based on a set of twenty-six text-based, primarily syntactic features and rank the relative importance of these features to identify a low-dimensional set still sufficient to achieve excellent classifier performance. This analysis demonstrates that Latin prose and verse can be classified effectively using just three top features. From examination of the highly ranked features, we observe that measures of the hypotactic style favored in Latin prose (i.e. subordinating constructions in complex sentences, such as relative clauses) are especially useful for classification.

@article{mcgillivray_computational_2019,
	title = {A computational approach to lexical polysemy in {Ancient} {Greek}},
	volume = {34},
	issn = {2055-7671},
	url = {https://doi.org/10.1093/llc/fqz036},
	doi = {10.1093/llc/fqz036},
	abstract = {Language is a complex and dynamic system. If we consider word meaning, which is the scope of lexical semantics, we observe that some words have several meanings, thus displaying lexical polysemy. In this article, we present the first phase of a project that aims at computationally modelling Ancient Greek semantics over time. Our system is based on Bayesian learning and on the Diorisis Ancient Greek corpus, which we have built for this purpose. We illustrate preliminary results in light of expert annotation, and take this opportunity to discuss the role of computational systems and human analysis in a complex research area like historical semantics. On the one hand, computational approaches allow us to model large corpora of texts. On the other hand, a long and rich scholarly tradition in Ancient Greek has provided us with valuable insights into the mechanisms of semantic change (cf. e.g. Leiwo, M. (2012). Introduction: variation with multiple faces. In Leiwo, M., Halla-aho, H., and Vierros, M. (eds), Variation and Change in Greek and Latin, Helsinki: Suomen Ateenan-instituutin säätiö, pp. 1–11.). In this article, we show that these qualitative analyses can be leveraged to support and complement the computational modelling.},
	number = {4},
	urldate = {2023-08-26},
	journal = {Digital Scholarship in the Humanities},
	author = {McGillivray, Barbara and Hengchen, Simon and Lähteenoja, Viivi and Palma, Marco and Vatri, Alessandro},
	month = dec,
	year = {2019},
	pages = {893--907},
}

Language is a complex and dynamic system. If we consider word meaning, which is the scope of lexical semantics, we observe that some words have several meanings, thus displaying lexical polysemy. In this article, we present the first phase of a project that aims at computationally modelling Ancient Greek semantics over time. Our system is based on Bayesian learning and on the Diorisis Ancient Greek corpus, which we have built for this purpose. We illustrate preliminary results in light of expert annotation, and take this opportunity to discuss the role of computational systems and human analysis in a complex research area like historical semantics. On the one hand, computational approaches allow us to model large corpora of texts. On the other hand, a long and rich scholarly tradition in Ancient Greek has provided us with valuable insights into the mechanisms of semantic change (cf. e.g. Leiwo, M. (2012). Introduction: variation with multiple faces. In Leiwo, M., Halla-aho, H., and Vierros, M. (eds), Variation and Change in Greek and Latin, Helsinki: Suomen Ateenan-instituutin säätiö, pp. 1–11.). In this article, we show that these qualitative analyses can be leveraged to support and complement the computational modelling.

@misc{nagy_metre_2019,
	title = {Metre as a stylometric feature in {Latin} hexameter poetry},
	url = {http://arxiv.org/abs/1911.12478},
	abstract = {This paper demonstrates that metre is a privileged indicator of authorial style in classical Latin hexameter poetry. Using only metrical features, pairwise classification experiments are performed between 5 first-century authors (10 comparisons) using four different machine-learning models. The results showed a two-label classification accuracy of at least 95\% with samples as small as ten lines and no greater than eighty lines (up to around 500 words). These sample sizes are an order of magnitude smaller than those typically recommended for BOW ('bag of words') or n-gram approaches, and the reported accuracy is outstanding. Additionally, this paper explores the potential for novelty (forgery) detection, or 'one-class classification'. An analysis of the disputed Aldine Additamentum (Sil. Ital. Puni. 8:144-225) concludes (p=0.0013) that the metrical style differs significantly from that of the rest of the poem.},
	language = {en},
	urldate = {2023-08-26},
	publisher = {arXiv},
	author = {Nagy, Benjamin},
	month = dec,
	year = {2019},
	note = {arXiv:1911.12478 [cs, stat]},
	keywords = {Computer Science - Computation and Language, Computer Science - Machine Learning, Statistics - Applications},
}

This paper demonstrates that metre is a privileged indicator of authorial style in classical Latin hexameter poetry. Using only metrical features, pairwise classification experiments are performed between 5 first-century authors (10 comparisons) using four different machine-learning models. The results showed a two-label classification accuracy of at least 95% with samples as small as ten lines and no greater than eighty lines (up to around 500 words). These sample sizes are an order of magnitude smaller than those typically recommended for BOW ('bag of words') or n-gram approaches, and the reported accuracy is outstanding. Additionally, this paper explores the potential for novelty (forgery) detection, or 'one-class classification'. An analysis of the disputed Aldine Additamentum (Sil. Ital. Puni. 8:144-225) concludes (p=0.0013) that the metrical style differs significantly from that of the rest of the poem.

@misc{assael_restoring_2019,
	title = {Restoring ancient text using deep learning: a case study on {Greek} epigraphy},
	shorttitle = {Restoring ancient text using deep learning},
	url = {http://arxiv.org/abs/1910.06262},
	abstract = {Ancient History relies on disciplines such as Epigraphy, the study of ancient inscribed texts, for evidence of the recorded past. However, these texts, “inscriptions”, are often damaged over the centuries, and illegible parts of the text must be restored by specialists, known as epigraphists. This work presents PYTHIA, the first ancient text restoration model that recovers missing characters from a damaged text input using deep neural networks. Its architecture is carefully designed to handle longterm context information, and deal efficiently with missing or corrupted character and word representations. To train it, we wrote a nontrivial pipeline to convert PHI, the largest digital corpus of ancient Greek inscriptions, to machine actionable text, which we call PHI-ML. On PHI-ML, PYTHIA’s predictions achieve a 30.1\% character error rate, compared to the 57.3\% of human epigraphists. Moreover, in 73.5\% of cases the ground-truth sequence was among the Top-20 hypotheses of PYTHIA, which effectively demonstrates the impact of this assistive method on the field of digital epigraphy, and sets the state-of-the-art in ancient text restoration.},
	language = {en},
	urldate = {2023-01-26},
	publisher = {arXiv},
	author = {Assael, Yannis and Sommerschield, Thea and Prag, Jonathan},
	month = oct,
	year = {2019},
	note = {arXiv:1910.06262 [cs]},
	keywords = {Computer Science - Computation and Language, Computer Science - Computers and Society},
}

Ancient History relies on disciplines such as Epigraphy, the study of ancient inscribed texts, for evidence of the recorded past. However, these texts, “inscriptions”, are often damaged over the centuries, and illegible parts of the text must be restored by specialists, known as epigraphists. This work presents PYTHIA, the first ancient text restoration model that recovers missing characters from a damaged text input using deep neural networks. Its architecture is carefully designed to handle longterm context information, and deal efficiently with missing or corrupted character and word representations. To train it, we wrote a nontrivial pipeline to convert PHI, the largest digital corpus of ancient Greek inscriptions, to machine actionable text, which we call PHI-ML. On PHI-ML, PYTHIA’s predictions achieve a 30.1% character error rate, compared to the 57.3% of human epigraphists. Moreover, in 73.5% of cases the ground-truth sequence was among the Top-20 hypotheses of PYTHIA, which effectively demonstrates the impact of this assistive method on the field of digital epigraphy, and sets the state-of-the-art in ancient text restoration.

@incollection{cayless_sustaining_2019,
	address = {Berlin},
	series = {Age of {Access}? {Grundfragen} der {Informationsgesellschaft}},
	title = {Sustaining {Linked} {Ancient} {World} {Data}},
	volume = {10},
	abstract = {Abstract: May 31st, 2018 marked the sixth anniversary of the Linked Ancient World Data Institute (LAWDI), a workshop funded by the US National Endowment For the Humanities. This makes it a good time to take stock of the Ancient World Linked Data initiatives that have been around for some time, as
well as some that have foundered and some that are new. What makes for sustainable Linked Open Data? Why do some initiatives thrive while others fail? What resources do successful LOD sites need, and how may they be obtained? The promise of LOD is that it frees our information from the silos in
which it is housed, permitting cross-system interactions that improve the quality and usefulness of the information in any single system. This article will take the broader view of the definition of Linked Data suggested by Tim Berners-Lee’s foundational “Linked Data – Design Issues” paper, as encompassing more types of data than simply RDF and other “Semantic Web” technologies. This view of LOD is pragmatic and leverages the strengths of semantic technologies while avoiding their weaknesses.},
	booktitle = {Digital classical philology: {Ancient} {Greek} and {Latin} in the digital revolution},
	publisher = {Walter de Gruyter \& Co},
	author = {Cayless, Hugh A.},
	year = {2019},
	pages = {35--50},
}

Abstract: May 31st, 2018 marked the sixth anniversary of the Linked Ancient World Data Institute (LAWDI), a workshop funded by the US National Endowment For the Humanities. This makes it a good time to take stock of the Ancient World Linked Data initiatives that have been around for some time, as well as some that have foundered and some that are new. What makes for sustainable Linked Open Data? Why do some initiatives thrive while others fail? What resources do successful LOD sites need, and how may they be obtained? The promise of LOD is that it frees our information from the silos in which it is housed, permitting cross-system interactions that improve the quality and usefulness of the information in any single system. This article will take the broader view of the definition of Linked Data suggested by Tim Berners-Lee’s foundational “Linked Data – Design Issues” paper, as encompassing more types of data than simply RDF and other “Semantic Web” technologies. This view of LOD is pragmatic and leverages the strengths of semantic technologies while avoiding their weaknesses.

@inproceedings{celano_standoff_2019,
	address = {New York, NY, USA},
	series = {{DATeCH2019}},
	title = {Standoff {Annotation} for the {Ancient} {Greek} and {Latin} {Dependency} {Treebank}},
	isbn = {978-1-4503-7194-0},
	url = {https://doi.org/10.1145/3322905.3322919},
	doi = {10.1145/3322905.3322919},
	abstract = {This contribution presents the work in progress to convert the Ancient Greek and Latin Dependency Treebank (AGLDT) into standoff annotation using PAULA XML. With an increasing number of annotations of any kind, it becomes more and more urgent that annotations related to the same texts be added standoff. Standoff annotation consists in adding any kind of annotation in separate documents, which are ultimately linked to a main text, the so-called "base text," which is meant to be unchangeable. References occur via a graph-based system of IDs, which allows an annotation layer (contained in a separate file) to be linked to another annotation layer (contained in another separate file). All the annotations/files create a labeled directed acyclic graph, whose root is represented by the base text. Standoff annotation enables easy interoperability and extension, in that single annotation layers can reference other layers of annotation independently, thus overcoming the problem of conflicting hierarchies. Moreover, standoff annotation also allows addition of different annotations of the same kind to the same text (e.g., two different interpretations of the POS tag for a given token). In the present contribution, I show how the annotations of the AGLDT can become standoff using PAULA XML, which is an open access format following the LAF principles. More precisely, I show the case study of Caesar's De Bello Civili. I detail the PAULA XML files created for its tokenization and sentence split, which are preliminary required to add morphosyntactic annotation.},
	urldate = {2023-04-26},
	booktitle = {Proceedings of the 3rd {International} {Conference} on {Digital} {Access} to {Textual} {Cultural} {Heritage}},
	publisher = {Association for Computing Machinery},
	author = {Celano, Giuseppe G. A.},
	year = {2019},
	keywords = {Ancient Greek, Latin, PAULA XML, dependency treebank, standoff annotation},
	pages = {149--153},
}

This contribution presents the work in progress to convert the Ancient Greek and Latin Dependency Treebank (AGLDT) into standoff annotation using PAULA XML. With an increasing number of annotations of any kind, it becomes more and more urgent that annotations related to the same texts be added standoff. Standoff annotation consists in adding any kind of annotation in separate documents, which are ultimately linked to a main text, the so-called "base text," which is meant to be unchangeable. References occur via a graph-based system of IDs, which allows an annotation layer (contained in a separate file) to be linked to another annotation layer (contained in another separate file). All the annotations/files create a labeled directed acyclic graph, whose root is represented by the base text. Standoff annotation enables easy interoperability and extension, in that single annotation layers can reference other layers of annotation independently, thus overcoming the problem of conflicting hierarchies. Moreover, standoff annotation also allows addition of different annotations of the same kind to the same text (e.g., two different interpretations of the POS tag for a given token). In the present contribution, I show how the annotations of the AGLDT can become standoff using PAULA XML, which is an open access format following the LAF principles. More precisely, I show the case study of Caesar's De Bello Civili. I detail the PAULA XML files created for its tokenization and sentence split, which are preliminary required to add morphosyntactic annotation.

@incollection{forstall_lexical_2019,
	address = {Cham},
	title = {Lexical {Matching}: {Text} {Reuse} as {Intertextuality}},
	isbn = {978-3-030-23415-7},
	shorttitle = {Lexical {Matching}},
	url = {https://doi.org/10.1007/978-3-030-23415-7_3},
	doi = {10.1007/978-3-030-23415-7_3},
	abstract = {Words are the basic unit for intertextual search in digital humanities. While published studies employ a variety of features depending on the textual practices with which they are concerned, the vast majority are combinations and transformations of word tokens. In this chapter, we introduce the idea of intertextual discovery through text-reuse detection, and present the core process common to most contemporary analyses.},
	language = {en},
	urldate = {2023-04-25},
	booktitle = {Quantitative {Intertextuality}: {Analyzing} the {Markers} of {Information} {Reuse}},
	publisher = {Springer International Publishing},
	author = {Forstall, Christopher W. and Scheirer, Walter J.},
	editor = {Forstall, Christopher W. and Scheirer, Walter J.},
	year = {2019},
	pages = {55--78},
}

Words are the basic unit for intertextual search in digital humanities. While published studies employ a variety of features depending on the textual practices with which they are concerned, the vast majority are combinations and transformations of word tokens. In this chapter, we introduce the idea of intertextual discovery through text-reuse detection, and present the core process common to most contemporary analyses.

@book{forstall_quantitative_2019,
	address = {Cham},
	title = {Quantitative {Intertextuality}: {Analyzing} the {Markers} of {Information} {Reuse}},
	isbn = {978-3-030-23413-3 978-3-030-23415-7},
	shorttitle = {Quantitative {Intertextuality}},
	url = {http://link.springer.com/10.1007/978-3-030-23415-7},
	doi = {10.1007/978-3-030-23415-7},
	language = {en},
	urldate = {2023-04-25},
	publisher = {Springer International Publishing},
	author = {Forstall, Christopher W. and Scheirer, Walter J.},
	year = {2019},
	keywords = {Authorship Attribution, Cultural Studies, Information Retrieval, Information Reuse, Pattern Matching},
}

@inproceedings{mambrini_linked_2019,
	address = {Paris, France},
	title = {Linked {Open} {Treebanks}. {Interlinking} {Syntactically} {Annotated} {Corpora} in the {LiLa} {Knowledge} {Base} of {Linguistic} {Resources} for {Latin}},
	abstract = {Paper about the procedure of inclusion of treebanks into the LiLa Knowledge Base of Linguistic Resources for Latin.},
	booktitle = {Proceedings of the 18th {International} {Workshop} on {Treebanks} and {Linguistic} {Theories} ({TLT}, {SyntaxFest} 2019)},
	author = {Mambrini, Francesco and Passarotti, Marco},
	year = {2019},
	keywords = {Latin, Linked Data, Treebank},
	pages = {74--81},
}

Paper about the procedure of inclusion of treebanks into the LiLa Knowledge Base of Linguistic Resources for Latin.

@inproceedings{ochab_stylometry_2019,
	address = {New York, NY, USA},
	series = {{DATeCH2019}},
	title = {Stylometry of literary papyri},
	isbn = {978-1-4503-7194-0},
	url = {https://doi.org/10.1145/3322905.3322930},
	doi = {10.1145/3322905.3322930},
	abstract = {In this paper we present the first results of stylometric analysis of literary papyri. Specifically we perform a range of tests for unsupervised clustering of authors. We scrutinise both the best classic distance-based methods as well as the state-of-the-art network community detection techniqes. We report on obstacles concerning highly non-uniform distributions of text size and authorial samples combined with sparse feature space. We also note how clustering performance depends on regularisation of spelling by means of querying relevant annotations.},
	urldate = {2023-04-25},
	booktitle = {Proceedings of the 3rd {International} {Conference} on {Digital} {Access} to {Textual} {Cultural} {Heritage}},
	publisher = {Association for Computing Machinery},
	author = {Ochab, Jeremi K. and Essler, Holger},
	year = {2019},
	keywords = {ancient Greek, authorship attribution, papyri, stylometry},
	pages = {139--142},
}

In this paper we present the first results of stylometric analysis of literary papyri. Specifically we perform a range of tests for unsupervised clustering of authors. We scrutinise both the best classic distance-based methods as well as the state-of-the-art network community detection techniqes. We report on obstacles concerning highly non-uniform distributions of text size and authorial samples combined with sparse feature space. We also note how clustering performance depends on regularisation of spelling by means of querying relevant annotations.

@article{koehler_interactive_2018,
	title = {Interactive {Classrooms} with {Jupyter} and {Python}},
	volume = {111},
	issn = {00255769},
	url = {http://www.jstor.org/stable/10.5951/mathteacher.111.4.0304},
	abstract = {A Jupyter notebook, a Web-based platform that teachers and students can use in a dynamic way, provides an interactive way to engage students' use of a computer to solve mathematical problems.},
	number = {4},
	urldate = {2026-01-25},
	journal = {The Mathematics Teacher},
	publisher = {National Council of Teachers of Mathematics},
	author = {Koehler, Jacob Frias and Kim, Soomi},
	year = {2018},
	pages = {304--308},
}

A Jupyter notebook, a Web-based platform that teachers and students can use in a dynamic way, provides an interactive way to engage students' use of a computer to solve mathematical problems.

@inproceedings{miranda_digital_2018,
	address = {Cham},
	title = {Digital {Literacy} in {Higher} {Education}},
	isbn = {978-3-319-91152-6},
	abstract = {The promotion of digital literacy is a priority for most information societies, as they become highly dependent on digital resources. As one of the core requirements of the workplace, digital literacy is increasingly being incorporated in higher education curricula. This paper aims to depict the level of digital literacy of higher education students, based on their own perceptions. The multiplicity of existing definitions for digital literacy constitutes a difficulty in its assessment. Hence, this paper begins by proposing a threefold definition of digital literacy: access to technology, operational competences and conceptual skills. This definition was deconstructed into sets of specific skills that were the subject of an online questionnaire that Portuguese higher education students were asked to complete. The analysis of the students’ self-evaluation in terms of their digital literacy, offered an important insight into their specific strengths and weaknesses when it comes to Information and Communication Technology.},
	booktitle = {Learning and {Collaboration} {Technologies}. {Learning} and {Teaching}},
	publisher = {Springer International Publishing},
	author = {Miranda, Paula and Isaias, Pedro and Pifano, Sara},
	editor = {Zaphiris, Panayiotis and Ioannou, Andri},
	year = {2018},
	pages = {71--87},
}

The promotion of digital literacy is a priority for most information societies, as they become highly dependent on digital resources. As one of the core requirements of the workplace, digital literacy is increasingly being incorporated in higher education curricula. This paper aims to depict the level of digital literacy of higher education students, based on their own perceptions. The multiplicity of existing definitions for digital literacy constitutes a difficulty in its assessment. Hence, this paper begins by proposing a threefold definition of digital literacy: access to technology, operational competences and conceptual skills. This definition was deconstructed into sets of specific skills that were the subject of an online questionnaire that Portuguese higher education students were asked to complete. The analysis of the students’ self-evaluation in terms of their digital literacy, offered an important insight into their specific strengths and weaknesses when it comes to Information and Communication Technology.

@techreport{high-level_expert_group_on_artificial_intelligence_definition_2018,
	address = {Brüssel},
	title = {A definition of {AI}: {Main} capabilities and scientific disciplines},
	url = {https://ec.europa.eu/futurium/en/system/files/ged/ai_hleg_definition_of_ai_18_december_1.pdf},
	language = {en},
	institution = {European Commission},
	author = {{High-Level Expert Group on Artificial Intelligence}},
	editor = {{European Commission}},
	month = dec,
	year = {2018},
}

@inproceedings{feng_improving_2018,
	address = {Stockholm, Sweden},
	title = {Improving {Low} {Resource} {Named} {Entity} {Recognition} using {Cross}-lingual {Knowledge} {Transfer}},
	isbn = {978-0-9992411-2-7},
	url = {https://www.ijcai.org/proceedings/2018/566},
	doi = {10.24963/ijcai.2018/566},
	abstract = {Neural networks have been widely used for high resource language (e.g. English) named entity recognition (NER) and have shown state-of-the-art results. However, for low resource languages, such as Dutch and Spanish, due to the limitation of resources and lack of annotated data, NER models tend to have lower performances. To narrow this gap, we investigate cross-lingual knowledge to enrich the semantic representations of low resource languages. We first develop neural networks to improve low resource word representations via knowledge transfer from high resource language using bilingual lexicons. Further, a lexicon extension strategy is designed to address out-of lexicon problem by automatically learning semantic projections. Finally, we regard word-level entity type distribution features as an external languageindependent knowledge and incorporate them into our neural architecture. Experiments on two low resource languages (Dutch and Spanish) demonstrate the effectiveness of these additional semantic representations (average 4.8\% improvement). Moreover, on Chinese OntoNotes 4.0 dataset, our approach achieves an F-score of 83.07\% with 2.91\% absolute gain compared to the state-of-the-art systems.},
	language = {en},
	urldate = {2025-01-01},
	booktitle = {Proceedings of the {Twenty}-{Seventh} {International} {Joint} {Conference} on {Artificial} {Intelligence}},
	publisher = {International Joint Conferences on Artificial Intelligence Organization},
	author = {Feng, Xiaocheng and Feng, Xiachong and Qin, Bing and Feng, Zhangyin and Liu, Ting},
	month = jul,
	year = {2018},
	pages = {4071--4077},
}

Neural networks have been widely used for high resource language (e.g. English) named entity recognition (NER) and have shown state-of-the-art results. However, for low resource languages, such as Dutch and Spanish, due to the limitation of resources and lack of annotated data, NER models tend to have lower performances. To narrow this gap, we investigate cross-lingual knowledge to enrich the semantic representations of low resource languages. We first develop neural networks to improve low resource word representations via knowledge transfer from high resource language using bilingual lexicons. Further, a lexicon extension strategy is designed to address out-of lexicon problem by automatically learning semantic projections. Finally, we regard word-level entity type distribution features as an external languageindependent knowledge and incorporate them into our neural architecture. Experiments on two low resource languages (Dutch and Spanish) demonstrate the effectiveness of these additional semantic representations (average 4.8% improvement). Moreover, on Chinese OntoNotes 4.0 dataset, our approach achieves an F-score of 83.07% with 2.91% absolute gain compared to the state-of-the-art systems.

@book{bengfort_applied_2018,
	address = {Boston},
	title = {Applied {Text} {Analysis} with {Python}: {Enabling} {Language}-{Aware} {Data} {Products} with {Machine} {Learning}.},
	shorttitle = {Applied {Text} {Analysis} with {Python}},
	url = {https://pdfroom.com/books/applied-text-analysis-with-python-enabling-language-aware-data-products-with-machine-learning/ra5179J6gJO},
	abstract = {Applied Text Analysis with Python: Enabling Language-Aware Data Products with Machine Learning},
	language = {en},
	urldate = {2023-12-31},
	publisher = {O'Reilly},
	author = {Bengfort, Benjamin and Ojeda, Tony and Bilbro, Rebecca},
	year = {2018},
}

Applied Text Analysis with Python: Enabling Language-Aware Data Products with Machine Learning

@article{eckhoff_proiel_2018,
	title = {The {PROIEL} treebank family: a standard for early attestations of {Indo}-{European} languages},
	volume = {52},
	doi = {https://doi.org/10.1007/s10579-017-9388-5},
	journal = {Lang Resources \& Evaluation},
	author = {Eckhoff, H. and Bech, K. and Bouma, G.},
	year = {2018},
	pages = {29--65},
}

@article{ponti_non-configurationality_2018,
	title = {Non-{Configurationality} in {Diachrony} : {Correlations} in {Local} and {Global} {Networks} of {Ancient} {Greek} and {Latin}},
	volume = {35},
	url = {doi:10.1075/dia.00007.pon.},
	number = {3},
	journal = {Diachronica : international journal for historical linguistics = revue internationale pour la linguistique historique = internationale Zeitschrift für historische Linguistik},
	author = {Ponti, Edoardo Maria and Luraghi, Silvia},
	year = {2018},
	pages = {367--392},
}

@incollection{cecchini_challenges_2018,
	address = {Bruxelles, Belgium},
	series = {Special {Interest} {Group} on linguistic {DATa} and corpus-based approaches to {NLP} ({SIGDAT}), {ACL}.},
	title = {Challenges in {Converting} the {Index} {Thomisticus} {Treebank} into {Universal} {Dependencies}},
	url = {https://www.aclweb.org/anthology/W18-6004.pdf},
	booktitle = {Proceedings of the {Second} {Workshop} on {Universal} {Dependencies} ({UDW} 2018) at {EMNLP} 2018},
	author = {Cecchini, F. M. and Passarotti, M. and Marongiu, P. and Zeman, D.},
	year = {2018},
	pages = {27--36},
}

@incollection{berti_annotating_2018,
	address = {Zürich/New York},
	title = {Annotating {Text} {Reuse} within the {Context}: {The} {Leipzig} {Open} {Fragmentary} {Texts} {Series} ({LOFTS})},
	language = {English},
	booktitle = {Text, {Kontext}, {Kontextualisierung}. {Moderne} {Kontextkonzepte} und antike {Literatur}},
	author = {Berti, Monica},
	editor = {Forst, Alexandra and Gärtner, Ursula and Tischer, Ute},
	year = {2018},
	pages = {223--234},
}

@inproceedings{monachini_digital_2018,
	title = {Digital {Classics}: {A} {Survey} on the {Needs} of {Ancient} {Greek} {Scholars} in {Italy}},
	url = {https://www.clarin.eu/sites/default/files/Monachini-Nicolosi-Stefanini-CLARIN2017_paper_3.pdf},
	abstract = {This paper presents and discusses the findings of a survey carried out in order to assess the use of digital resources and digital technologies with respect to work in Ancient Greek scholarship, as well as to identify the factors that are likely to constrain its use and to elicit needs and requirements of Ancient Greek scholars in Italy. The survey is in line with the principles behind the recent user engagement strategy developed by CLARIN-ERIC and constitutes one of the national efforts undertaken by CLARIN-IT to contribute to the wider impact of CLARIN on Digital Classicists.},
	language = {en},
	booktitle = {Proceedings of the {CLARIN} 2017 {Conference}},
	publisher = {Linköping University Electronic Press},
	author = {Monachini, Monica and Nicolosi, Anika and Stefanini, Alberto},
	year = {2018},
	keywords = {reference},
	pages = {1--5},
}

This paper presents and discusses the findings of a survey carried out in order to assess the use of digital resources and digital technologies with respect to work in Ancient Greek scholarship, as well as to identify the factors that are likely to constrain its use and to elicit needs and requirements of Ancient Greek scholars in Italy. The survey is in line with the principles behind the recent user engagement strategy developed by CLARIN-ERIC and constitutes one of the national efforts undertaken by CLARIN-IT to contribute to the wider impact of CLARIN on Digital Classicists.

@phdthesis{nury_automated_2018,
	type = {phdthesis},
	title = {Automated {Collation} and {Digital} {Editions}: from {Theory} to {Practice}},
	shorttitle = {Automated {Collation} and {Digital} {Editions}},
	url = {https://hal.science/tel-02493805},
	abstract = {The purpose of the dissertation is to investigate from a theoretical and methodological perspective the different tools that allow automated collation, and study the application of such tools to the creation of a digital critical edition in the context of Classical literature. By doing so, the dissertation examines many foundational but often neglected components of the philological method, such as the definition and wider implication of transcription, reading, and variant. The goal is to provide a reflection on automated collation and the theoretical as well as practical challenges it poses: what is automated collation? How is it performed, and what are the main differences with manual collation? What are the benefits of automated collation? Why has it not been widely adopted yet, despite the fact that it was developed to help scholars? How to process the results of collation programmes? As a case study, a Classical Latin text has been used to test automated collation and to compare the various existing tools.},
	language = {en},
	urldate = {2023-08-26},
	school = {King's College London},
	author = {Nury, Elisa},
	month = jul,
	year = {2018},
}

The purpose of the dissertation is to investigate from a theoretical and methodological perspective the different tools that allow automated collation, and study the application of such tools to the creation of a digital critical edition in the context of Classical literature. By doing so, the dissertation examines many foundational but often neglected components of the philological method, such as the definition and wider implication of transcription, reading, and variant. The goal is to provide a reflection on automated collation and the theoretical as well as practical challenges it poses: what is automated collation? How is it performed, and what are the main differences with manual collation? What are the benefits of automated collation? Why has it not been widely adopted yet, despite the fact that it was developed to help scholars? How to process the results of collation programmes? As a case study, a Classical Latin text has been used to test automated collation and to compare the various existing tools.

@inproceedings{lana_eliciting_2018,
	address = {Cham},
	series = {Communications in {Computer} and {Information} {Science}},
	title = {Eliciting the {Ancient} {Geography} from a {Digital} {Library} of {Latin} {Texts}},
	isbn = {978-3-319-73165-0},
	doi = {10.1007/978-3-319-73165-0_19},
	abstract = {Geolat – Geography for Latin Literature is a research project, aimed at making accessible a digital library containing the works of Latin literature (from its origins in 240 BCE to the end of the Roman Empire in 476 CE) through a query interface of geographic/cartographic type representing the geographic knowledge expressed in the Latin texts themselves. A core activity of the project has been the development of the ontology GO!, which describes the geographical knowledge contained in the texts of the library. The ontologically annotated texts will allow for a variety of scientifically relevant uses, apart from the geo-based browsing: for example the production of digital and printed critical editions. The project is under development at Dipartimento di Studi Umanistici of Università del Piemonte Orientale, and financially supported by Fondazione Compagnia di San Paolo.},
	language = {en},
	booktitle = {Digital {Libraries} and {Multimedia} {Archives}},
	publisher = {Springer International Publishing},
	author = {Lana, Maurizio and Tambassi, Timothy},
	editor = {Serra, Giuseppe and Tasso, Carlo},
	year = {2018},
	keywords = {Classical latin texts, Digital library, Geography, OWL, Ontology, Web},
	pages = {191--200},
}

Geolat – Geography for Latin Literature is a research project, aimed at making accessible a digital library containing the works of Latin literature (from its origins in 240 BCE to the end of the Roman Empire in 476 CE) through a query interface of geographic/cartographic type representing the geographic knowledge expressed in the Latin texts themselves. A core activity of the project has been the development of the ontology GO!, which describes the geographical knowledge contained in the texts of the library. The ontologically annotated texts will allow for a variety of scientifically relevant uses, apart from the geo-based browsing: for example the production of digital and printed critical editions. The project is under development at Dipartimento di Studi Umanistici of Università del Piemonte Orientale, and financially supported by Fondazione Compagnia di San Paolo.

@article{zhang_did_2018,
	title = {Did {Gaius} {Julius} {Caesar} {Write} {De} {Bello} {Hispaniensi}? {A} {Computational} {Study} of {Latin} {Classics} {Authorship}},
	volume = {14},
	copyright = {Copyright (c) 2018 Human IT: Journal for Information Technology Studies as a Human Science},
	issn = {1402-151X},
	shorttitle = {Did {Gaius} {Julius} {Caesar} {Write} {De} {Bello} {Hispaniensi}?},
	url = {https://humanit.hb.se/article/view/515},
	abstract = {This project addresses a two-millennium old mystery surrounding the authorship of ancient Latin war memoirs attributed to Caesar, using Distributional Semantics, a modern computational method for detecting written text patterns. The Civil War has been confirmed to be Caesar’s work, as well as the first seven of the eight chapters of the Gallic War, the eighth by Hirtius. The authorship of the African, Alexandrine, and Spanish Wars, though attributed to Caesar, is still under debate. Methods of distributional semantics derive representations of words from their distribution across a large amount of text, such that words that occur in similar contexts have similar representations.  These representations can then be combined to model larger units of text, such as chapters and whole books. SemanticVectors software was used to calculate the similarity between chapters or books after dimension reduction using Random Indexing. The results show that the Gallic War’s eighth chapter is significantly different from its other seven chapters and from the Civil War, verifying the ability of distributional semantics to detect different Latin authorships. The African, Alexandrine, and Spanish Wars are notably different from the Civil War andGallic War (first seven chapters), suggesting that Caesar did not write these three. Furthermore, the African, Alexandrine, and Spanish Wars are different from each other and from the Civil and Gallic Wars, suggesting that they were written by different authors. This project demonstrates the value of distributional semantics in classics research. Its implications for digital humanities and real world problems such as plagiarism are discussed.},
	language = {en},
	number = {1},
	urldate = {2023-08-26},
	journal = {Human IT: Journal for Information Technology Studies as a Human Science},
	author = {Zhang, Olivia R. and Cohen, Trevor and McGill, Scott},
	month = jun,
	year = {2018},
	note = {Number: 1},
	keywords = {Caesar, Classics, Latin, authorship attribution, computational linguistics, distributional semantics},
	pages = {28--58},
}

This project addresses a two-millennium old mystery surrounding the authorship of ancient Latin war memoirs attributed to Caesar, using Distributional Semantics, a modern computational method for detecting written text patterns. The Civil War has been confirmed to be Caesar’s work, as well as the first seven of the eight chapters of the Gallic War, the eighth by Hirtius. The authorship of the African, Alexandrine, and Spanish Wars, though attributed to Caesar, is still under debate. Methods of distributional semantics derive representations of words from their distribution across a large amount of text, such that words that occur in similar contexts have similar representations.  These representations can then be combined to model larger units of text, such as chapters and whole books. SemanticVectors software was used to calculate the similarity between chapters or books after dimension reduction using Random Indexing. The results show that the Gallic War’s eighth chapter is significantly different from its other seven chapters and from the Civil War, verifying the ability of distributional semantics to detect different Latin authorships. The African, Alexandrine, and Spanish Wars are notably different from the Civil War andGallic War (first seven chapters), suggesting that Caesar did not write these three. Furthermore, the African, Alexandrine, and Spanish Wars are different from each other and from the Civil and Gallic Wars, suggesting that they were written by different authors. This project demonstrates the value of distributional semantics in classics research. Its implications for digital humanities and real world problems such as plagiarism are discussed.

@article{healy_bridging_2017,
	title = {Bridging the {Cyber}-{Analysis} {Gap}: {The} {Democratization} of {Data} {Science}},
	volume = {2},
	issn = {24742120, 24742139},
	url = {http://www.jstor.org/stable/26267404},
	abstract = {ABSTRACT The challenges of ever growing and ever changing Big Data are broad and far-reaching, particularly in the cyber-defense domain. The task of analyzing and making sense of this data is difficult, and only getting worse. We propose that by democratizing data science and making it accessible to everyone, we can expand the breadth and depth of analytics available to a point where we can potentially meet the challenges of Big Data.},
	number = {1},
	urldate = {2026-01-25},
	journal = {The Cyber Defense Review},
	publisher = {Army Cyber Institute},
	author = {Healy, John and McInnes, Leland and Weir, Colin},
	year = {2017},
	pages = {109--118},
}

ABSTRACT The challenges of ever growing and ever changing Big Data are broad and far-reaching, particularly in the cyber-defense domain. The task of analyzing and making sense of this data is difficult, and only getting worse. We propose that by democratizing data science and making it accessible to everyone, we can expand the breadth and depth of analytics available to a point where we can potentially meet the challenges of Big Data.

@article{wautelet_user-story_2017,
	title = {User-story driven development of multi-agent systems: {A} process fragment for agile methods},
	volume = {50},
	issn = {1477-8424},
	shorttitle = {User-story driven development of multi-agent systems},
	url = {https://www.sciencedirect.com/science/article/pii/S1477842416301178},
	doi = {10.1016/j.cl.2017.06.007},
	abstract = {Agile software development methods are mostly built as a set of managerial guidelines and development concepts on how to handle a software development but are not bounded to software development paradigms like object or agent orientation. Some methods, like eXtreme Programming and SCRUM are driven by operational requirements representation models called User Stories. These User Stories can be used as an anchoring point to agile methods; this means that we could take a User Stories set to drive a software transformation approach embedded in a particular development paradigm. This paper presents a process fragment for Multi-Agent Systems development with agile methods based on User Stories sets. The process fragment indeed takes advantage of an initial set of User Stories to build a reasoning model (called the Rationale Tree; typically several of these are built for a single project) that documents decompositions and means-end alternatives in scenarios for requirements realization. A Rationale Tree can then be aligned with a Multi-Agent design and implemented in an agent-oriented development language. In this paper the transformation is targeted to the JAVA Agent DEvelopment (JADE) framework. The process fragment (at least partially) covers the Requirements Analysis, Multi-Agent System Design and Multi-Agent System Implementation phases. Transformation from one phase to the other is overseen and illustrated on an example.},
	urldate = {2025-12-13},
	journal = {Computer Languages, Systems \& Structures},
	author = {Wautelet, Yves and Heng, Samedi and Kiv, Soreangsey and Kolp, Manuel},
	month = dec,
	year = {2017},
	keywords = {Agent software engineering, Agile development, JADE, JAVA Agent DEvelopment framework, Multi-agent system, Process fragment, Rationale tree, User story, i*-based software process modeling},
	pages = {159--176},
}

Agile software development methods are mostly built as a set of managerial guidelines and development concepts on how to handle a software development but are not bounded to software development paradigms like object or agent orientation. Some methods, like eXtreme Programming and SCRUM are driven by operational requirements representation models called User Stories. These User Stories can be used as an anchoring point to agile methods; this means that we could take a User Stories set to drive a software transformation approach embedded in a particular development paradigm. This paper presents a process fragment for Multi-Agent Systems development with agile methods based on User Stories sets. The process fragment indeed takes advantage of an initial set of User Stories to build a reasoning model (called the Rationale Tree; typically several of these are built for a single project) that documents decompositions and means-end alternatives in scenarios for requirements realization. A Rationale Tree can then be aligned with a Multi-Agent design and implemented in an agent-oriented development language. In this paper the transformation is targeted to the JAVA Agent DEvelopment (JADE) framework. The process fragment (at least partially) covers the Requirements Analysis, Multi-Agent System Design and Multi-Agent System Implementation phases. Transformation from one phase to the other is overseen and illustrated on an example.

@book{carretero_digcomp_2017,
	address = {Luxembourg},
	title = {{DigComp} 2.1 - {The} {Digital} {Competence} {Framework} for {Citizens}},
	isbn = {978-92-79-68006-9},
	url = {https://op.europa.eu/en/publication-detail/-/publication/3c5e7879-308f-11e7-9412-01aa75ed71a1/language-en},
	abstract = {The European Digital Competence Framework for Citizens, also known as DigComp, offers a tool to improve citizens' digital competence. DigComp was first published in 2013 and has become a reference for many digital competence initiatives at both European and Member State levels. This document introduces DigComp 2.0. It constitutes phase 1 of the update of the framework which focuses on the conceptual reference model, new vocabulary and streamlined descriptors. The current document also gives examples of how DigComp is used at the European, national and regional levels.},
	publisher = {Publications Office},
	author = {Carretero, Stephanie and Vuorikari, Riina and Punie, Yves},
	year = {2017},
	note = {Backup Publisher: European Commission},
	keywords = {Community support framework, EU Member State, Education policy, European Union, Information technology and telecommunications, digital literacy, digital technology, impact of information technology, information technology, learning technique, new educational methods, new technology, report, research report},
}

The European Digital Competence Framework for Citizens, also known as DigComp, offers a tool to improve citizens' digital competence. DigComp was first published in 2013 and has become a reference for many digital competence initiatives at both European and Member State levels. This document introduces DigComp 2.0. It constitutes phase 1 of the update of the framework which focuses on the conceptual reference model, new vocabulary and streamlined descriptors. The current document also gives examples of how DigComp is used at the European, national and regional levels.

@article{vierros_preprocessing_2017,
	title = {Preprocessing {Greek} {Papyri} for {Linguistic} {Annotation}},
	doi = {https://doi.org/10.46298/jdmdh.1385},
	number = {Numéro spécial sur le traitement assisté par ordinateur de l‘intertextualité dans les langues anciennes},
	journal = {Journal of Data Mining \& Digital Humanities},
	author = {Vierros, Marja and Henriksson, Erik},
	year = {2017},
}

@book{mistrik_software_2017,
	title = {Software {Architecture} for {Big} {Data} and the {Cloud}},
	isbn = {978-0-12-809338-2},
	url = {https://books.google.de/books?hl=de&lr=&id=zvPtDQAAQBAJ},
	abstract = {Software Architecture for Big Data and the Cloud is designed to be a single resource that brings together research on how software architectures can solve the challenges imposed by building big data software systems. The challenges of big data on the software architecture can relate to scale, security, integrity, performance, concurrency, parallelism, and dependability, amongst others. Big data handling requires rethinking architectural solutions to meet functional and non-functional requirements related to volume, variety and velocity. The book's editors have varied and complementary backgrounds in requirements and architecture, specifically in software architectures for cloud and big data, as well as expertise in software engineering for cloud and big data. This book brings together work across different disciplines in software engineering, including work expanded from conference tracks and workshops led by the editors. Discusses systematic and disciplined approaches to building software architectures for cloud and big data with state-of-the-art methods and techniques Presents case studies involving enterprise, business, and government service deployment of big data applications Shares guidance on theory, frameworks, methodologies, and architecture for cloud and big data},
	language = {en},
	publisher = {Morgan Kaufmann},
	author = {Mistrik, Ivan and Bahsoon, Rami and Ali, Nour and Heisel, Maritta and Maxim, Bruce},
	month = jun,
	year = {2017},
	note = {Google-Books-ID: zvPtDQAAQBAJ},
	keywords = {Computers / Software Development \& Engineering / General},
}

Software Architecture for Big Data and the Cloud is designed to be a single resource that brings together research on how software architectures can solve the challenges imposed by building big data software systems. The challenges of big data on the software architecture can relate to scale, security, integrity, performance, concurrency, parallelism, and dependability, amongst others. Big data handling requires rethinking architectural solutions to meet functional and non-functional requirements related to volume, variety and velocity. The book's editors have varied and complementary backgrounds in requirements and architecture, specifically in software architectures for cloud and big data, as well as expertise in software engineering for cloud and big data. This book brings together work across different disciplines in software engineering, including work expanded from conference tracks and workshops led by the editors. Discusses systematic and disciplined approaches to building software architectures for cloud and big data with state-of-the-art methods and techniques Presents case studies involving enterprise, business, and government service deployment of big data applications Shares guidance on theory, frameworks, methodologies, and architecture for cloud and big data

@article{bodard_standards_2017,
	title = {Standards for {Networking} {Ancient} {Person} data: {Digital} approaches to problems in prosopographical space},
	copyright = {Copyright (c) 2017},
	issn = {2364-7957},
	shorttitle = {Standards for {Networking} {Ancient} {Person} data},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/37975},
	doi = {10.11588/dco.2017.0.37975},
	abstract = {Prosopographies disambiguate names appearing in sources by creating lists of persons, but the progress of scholarship now makes these lists difficult to maintain. In a digital context unique stable identifiers can be reshuffled ad libitum when searching and ordering information. Digital data increasingly brings together complementary research outputs: the Standards for Networking Ancient Prosopographies project takes on the challenge of creating an aggregated resource, adopting a Linked Open Data approach. In this paper we shall present three case studies highlighting the promise and problems of encoding unambiguous identities, titulature and other disambiguating information, and treating divine figures as person-data, respectively. Digital approaches are tools for research, assisting rather than replacing the historian, who remains central to the research endeavor.},
	language = {en},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Bodard, Gabriel and Cayless, Hugh and Depauw, Mark and Isaksen, Leif and Lawrence, Faith and Rahtz, Sebastian},
	month = nov,
	year = {2017},
	pages = {28--43},
}

Prosopographies disambiguate names appearing in sources by creating lists of persons, but the progress of scholarship now makes these lists difficult to maintain. In a digital context unique stable identifiers can be reshuffled ad libitum when searching and ordering information. Digital data increasingly brings together complementary research outputs: the Standards for Networking Ancient Prosopographies project takes on the challenge of creating an aggregated resource, adopting a Linked Open Data approach. In this paper we shall present three case studies highlighting the promise and problems of encoding unambiguous identities, titulature and other disambiguating information, and treating divine figures as person-data, respectively. Digital approaches are tools for research, assisting rather than replacing the historian, who remains central to the research endeavor.

@article{pockelmann_paraphrasensuche_2017,
	title = {Paraphrasensuche mittels word2vec und der {Word} {Mover}’s {Distance} im {Altgriechischen}},
	copyright = {Copyright (c) 2017 Digital Classics Online},
	issn = {2364-7957},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/40185},
	doi = {10.11588/dco.2017.0.40185},
	abstract = {To find receptions of Plato‘s work within the ancient Greek literature, automatic methods would be a useful assistance. Unfortunately, such methods are often knowledge-based and thus restricted to extensively annotated texts, which are not available to a sufficient extent for ancient Greek. In this article, we describe an approach that is based on the distributional hypotheses instead, to overcome the problem of missing annotations. This approach uses word2vec and the related Word Mover‘s Distance to determine phrases with similar meaning. Despite its experimental state, the method produces some meaningful results as shown in three examples.},
	language = {de},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Pöckelmann, Marcus and Ritter, Jörg and Wöckener-Gade, Eva and Schubert, Charlotte},
	month = dec,
	year = {2017},
	keywords = {Word Mover’s Distance},
	pages = {24--36},
}

To find receptions of Plato‘s work within the ancient Greek literature, automatic methods would be a useful assistance. Unfortunately, such methods are often knowledge-based and thus restricted to extensively annotated texts, which are not available to a sufficient extent for ancient Greek. In this article, we describe an approach that is based on the distributional hypotheses instead, to overcome the problem of missing annotations. This approach uses word2vec and the related Word Mover‘s Distance to determine phrases with similar meaning. Despite its experimental state, the method produces some meaningful results as shown in three examples.

@article{chaudhuri_bioinformatics_2017,
	title = {Bioinformatics and {Classical} {Literary} {Study}},
	volume = {Numéro spécial sur le traitement assisté par ordinateur de l‘intertextualité dans les langues anciennes},
	issn = {2416-5999},
	url = {https://jdmdh.episciences.org/3807},
	doi = {10.46298/jdmdh.1386},
	abstract = {This paper describes the Quantitative Criticism Lab, a collaborative initiative between classicists, quantitative biologists, and computer scientists to apply ideas and methods drawn from the sciences to the study of literature. A core goal of the project is the use of computational biology, natural language processing, and machine learning techniques to investigate authorial style, intertextuality, and related phenomena of literary significance. As a case study in our approach, here we review the use of sequence alignment, a common technique in genomics and computational linguistics, to detect intertextuality in Latin literature. Sequence alignment is distinguished by its ability to find inexact verbal similarities, which makes it ideal for identifying phonetic echoes in large corpora of Latin texts. Although especially suited to Latin, sequence alignment in principle can be extended to many other languages.},
	number = {Project presentations},
	urldate = {2023-08-26},
	journal = {Journal of Data Mining \& Digital Humanities},
	publisher = {Episciences.org},
	author = {Chaudhuri, Pramit and Dexter, Joseph P.},
	month = aug,
	year = {2017},
}

This paper describes the Quantitative Criticism Lab, a collaborative initiative between classicists, quantitative biologists, and computer scientists to apply ideas and methods drawn from the sciences to the study of literature. A core goal of the project is the use of computational biology, natural language processing, and machine learning techniques to investigate authorial style, intertextuality, and related phenomena of literary significance. As a case study in our approach, here we review the use of sequence alignment, a common technique in genomics and computational linguistics, to detect intertextuality in Latin literature. Sequence alignment is distinguished by its ability to find inexact verbal similarities, which makes it ideal for identifying phonetic echoes in large corpora of Latin texts. Although especially suited to Latin, sequence alignment in principle can be extended to many other languages.

@article{dexter_quantitative_2017,
	title = {Quantitative criticism of literary relationships},
	volume = {114},
	url = {https://www.pnas.org/doi/abs/10.1073/pnas.1611910114},
	doi = {10.1073/pnas.1611910114},
	abstract = {Authors often convey meaning by referring to or imitating prior works of literature, a process that creates complex networks of literary relationships (“intertextuality”) and contributes to cultural evolution. In this paper, we use techniques from stylometry and machine learning to address subjective literary critical questions about Latin literature, a corpus marked by an extraordinary concentration of intertextuality. Our work, which we term “quantitative criticism,” focuses on case studies involving two influential Roman authors, the playwright Seneca and the historian Livy. We find that four plays related to but distinct from Seneca’s main writings are differentiated from the rest of the corpus by subtle but important stylistic features. We offer literary interpretations of the significance of these anomalies, providing quantitative data in support of hypotheses about the use of unusual formal features and the interplay between sound and meaning. The second part of the paper describes a machine-learning approach to the identification and analysis of citational material that Livy loosely appropriated from earlier sources. We extend our approach to map the stylistic topography of Latin prose, identifying the writings of Caesar and his near-contemporary Livy as an inflection point in the development of Latin prose style. In total, our results reflect the integration of computational and humanistic methods to investigate a diverse range of literary questions.},
	number = {16},
	urldate = {2023-08-26},
	journal = {Proceedings of the National Academy of Sciences},
	publisher = {Proceedings of the National Academy of Sciences},
	author = {Dexter, Joseph P. and Katz, Theodore and Tripuraneni, Nilesh and Dasgupta, Tathagata and Kannan, Ajay and Brofos, James A. and Bonilla Lopez, Jorge A. and Schroeder, Lea A. and Casarez, Adriana and Rabinovich, Maxim and Haimson Lushkov, Ayelet and Chaudhuri, Pramit},
	month = apr,
	year = {2017},
	pages = {E3195--E3204},
}

Authors often convey meaning by referring to or imitating prior works of literature, a process that creates complex networks of literary relationships (“intertextuality”) and contributes to cultural evolution. In this paper, we use techniques from stylometry and machine learning to address subjective literary critical questions about Latin literature, a corpus marked by an extraordinary concentration of intertextuality. Our work, which we term “quantitative criticism,” focuses on case studies involving two influential Roman authors, the playwright Seneca and the historian Livy. We find that four plays related to but distinct from Seneca’s main writings are differentiated from the rest of the corpus by subtle but important stylistic features. We offer literary interpretations of the significance of these anomalies, providing quantitative data in support of hypotheses about the use of unusual formal features and the interplay between sound and meaning. The second part of the paper describes a machine-learning approach to the identification and analysis of citational material that Livy loosely appropriated from earlier sources. We extend our approach to map the stylistic topography of Latin prose, identifying the writings of Caesar and his near-contemporary Livy as an inflection point in the development of Latin prose style. In total, our results reflect the integration of computational and humanistic methods to investigate a diverse range of literary questions.

@article{almas_perseids_2017,
	title = {Perseids: {Experimenting} with {Infrastructure} for {Creating} and {Sharing} {Research} {Data} in the {Digital} {Humanities}},
	volume = {16},
	number = {19},
	journal = {Data Science Journal},
	author = {Almas, B.},
	year = {2017},
	pages = {1--17},
}

@inproceedings{kluyver_jupyter_2016,
	title = {Jupyter {Notebooks} - a publishing format for reproducible computational workflows},
	url = {https://api.semanticscholar.org/CorpusID:36928206},
	booktitle = {International {Conference} on {Electronic} {Publishing}},
	author = {Kluyver, Thomas and Ragan-Kelley, Benjamin and Pérez, Fernando and Granger, Brian E. and Bussonnier, Matthias and Frederic, Jonathan and Kelley, Kyle and Hamrick, Jessica B. and Grout, Jason and Corlay, Sylvain and Ivanov, Paul and Avila, Damián and Abdalla, Safia and Willing, Carol and Team, Jupyter Development},
	year = {2016},
}

@article{asper_representing_2016,
	title = {Representing {Authority} in {Ancient} {Knowledge} {Texts}},
	volume = {6: Space and Knowledge. Topoi Research Group Articles},
	url = {https://refubium.fu-berlin.de/bitstream/handle/fub188/21825/Asperx_Markus_Representing.pdf?sequence=1},
	doi = {10.17169/FUDOCS_document_000000026007},
	journal = {eTopoi Journal for Ancient Studies},
	author = {Asper, Markus and Renger, Almut-Barbara and Sala, Tudor and Witte, Markus and Pilhofer, Philipp and Walter, Sarah and Kirk, Nalini},
	year = {2016},
	pages = {389--417},
}

@book{sekretariat_der_kultusministerkonferenz_strategie_2016,
	address = {Berlin},
	title = {Strategie der {Kultusministerkonferenz} {Bildung} in der digitalen {Welt}},
	url = {https://www.kmk.org/fileadmin/Dateien/pdf/PresseUndAktuelles/2016/Bildung_digitale_Welt_Webversion.pdf},
	publisher = {KMK},
	editor = {{Sekretariat der Kultusministerkonferenz}},
	month = aug,
	year = {2016},
}

@incollection{romanello_introduction_2016,
	series = {Teaching, {Knowledge} {Exchange} \& {Public} {Engagement}},
	title = {Introduction},
	isbn = {978-1-909188-48-8},
	url = {https://www.jstor.org/stable/j.ctv3s8tgt.5},
	abstract = {Situated within the broader field of Digital Humanities, Digital Classics is concerned with the application of computational methods and theories to the study of the Greco-Roman and wider ancient world. Over the last decade or so, a decentralised and international community of researchers in this area has emerged, centred around the \textit{Digital Classicist}  . In addition to curating a wiki, and conversations over discussion lists in two languages, this community has been organising several seminar series aimed at providing a venue for discussion of work in progress. Indeed, some of the chapters in this collection arose from papers given at the},
	urldate = {2025-05-21},
	booktitle = {Digital {Classics} {Outside} the {Echo}-{Chamber}},
	publisher = {Ubiquity Press},
	author = {Romanello, Matteo and Bodard, Gabriel},
	editor = {Romanello, Matteo and Bodard, Gabriel},
	year = {2016},
	pages = {1--12},
}

Situated within the broader field of Digital Humanities, Digital Classics is concerned with the application of computational methods and theories to the study of the Greco-Roman and wider ancient world. Over the last decade or so, a decentralised and international community of researchers in this area has emerged, centred around the Digital Classicist . In addition to curating a wiki, and conversations over discussion lists in two languages, this community has been organising several seminar series aimed at providing a venue for discussion of work in progress. Indeed, some of the chapters in this collection arose from papers given at the

@inproceedings{erdmann_challenges_2016,
	title = {Challenges and solutions for {Latin} named entity recognition},
	url = {https://aclanthology.org/W16-4012.pdf},
	author = {Erdmann, Alexander and Brown, Christopher and Joseph, Brian and Janse, Mark and Ajaka, Petra and Elsner, Micha and de Marneffe, Marie-Catherine},
	year = {2016},
	pages = {85--93},
}

@article{dressler_nominal_2016,
	title = {Nominal {Intelligence}: {Conspiracy}, {Prosopography}, and the {Secret} of {Horace}},
	volume = {36},
	issn = {3110475871},
	url = {https://www.degruyter.com/document/doi/10.1515/9783110475876/pdf},
	journal = {Wordplay and Powerplay in Latin Poetry},
	publisher = {Walter de Gruyter GmbH \& Co KG},
	author = {Dressler, Alex},
	year = {2016},
	pages = {37--68},
}

@inproceedings{aguilar_named_2016,
	title = {Named entity recognition applied on a data base of {Medieval} {Latin} charters. {The} case of chartae burgundiae},
	url = {https://hal.science/hal-02407159/document},
	author = {Aguilar, Sergio Torres and Tannier, Xavier and Chastang, Pierre},
	year = {2016},
}

@inproceedings{torres_aguilar_named_2016,
	address = {Krakow},
	title = {Named entity recognition applied on a data base of {Medieval} {Latin} charters. {The} case of chartae burgundiae.},
	volume = {1632},
	isbn = {1613-0073},
	url = {https://ceur-ws.org/Vol-1632/paper_9.pdf},
	abstract = {The work on the named entity recognition (NER) in databases of historical texts has been placed among the most promising new ways to implement best recovery and managements tools for exploring mass data. In this paper, we describe the application processing NER through a modelling with CRF on an annotated database of Burgundy collection of charters from the tenth to thirteenth centuries. The aim is to generate a model for automatic recognition of named entities in historical sources. We discuss the nature of historical documents in the corpus and extraction of rules, and we expose adaptation to the processing algorithm and the most common problems encountered in Medio Latin texts using diplomatic formularies, which is an atypical case within the NER studies.},
	language = {en},
	booktitle = {Proceedings of the 3rd {HistoInformatics} {Workshop}},
	author = {Torres Aguilar, Sergio Octavio and Tannier, Xavier and Chastang, Pierre},
	year = {2016},
	pages = {67--71},
}

The work on the named entity recognition (NER) in databases of historical texts has been placed among the most promising new ways to implement best recovery and managements tools for exploring mass data. In this paper, we describe the application processing NER through a modelling with CRF on an annotated database of Burgundy collection of charters from the tenth to thirteenth centuries. The aim is to generate a model for automatic recognition of named entities in historical sources. We discuss the nature of historical documents in the corpus and extraction of rules, and we expose adaptation to the processing algorithm and the most common problems encountered in Medio Latin texts using diplomatic formularies, which is an atypical case within the NER studies.

@article{springmann_latmor_2016,
	title = {{LatMor} : {A} {Latin} {Finite}-{State} {Morphology} {Encoding} {Vowel} {Quantity}},
	volume = {2},
	url = {doi:10.1515/opli-2016-0019.},
	number = {1},
	journal = {Open linguistics},
	author = {Springmann, Uwe and Schmid, Helmut and Najock, Dietmar},
	year = {2016},
	pages = {386--392},
}

@book{korkiakangas_subject_2016,
	address = {Helsinki},
	series = {Commentationes {Humanarum} {Litterarum}},
	title = {Subject {Case} in the {Latin} of {Tuscan} {Charters} of the 8th and 9th {Centuries}.},
	volume = {133},
	publisher = {Societas Scientiarum Fennica},
	author = {Korkiakangas, Timo},
	year = {2016},
}

@incollection{mahony_open_2016,
	title = {Open {Education} and {Open} {Educational} {Resources} for the {Teaching} of {Classics} in the {UK}},
	url = {https://www.ubiquitypress.com/site/books/10.5334/bat/},
	doi = {10.5334/bat},
	abstract = {Edited by organizers of “Digital Classicist” seminars in London and Berlin, this volume addresses the impact of computational approaches to the study of antiquity on audiences other than the scholars who conventionally publish it.  In addition to colleagues in classics and digital humanities, the eleven chapters herein concern and are addressed to students, heritage professionals and “citizen scientists”.{\textless}br{\textgreater}{\textless}br{\textgreater}Each chapter is a scholarly contribution, presenting research questions in the classics, digital humanities or, in many cases, both. They are all also examples of work within one of the most important areas of academia today: scholarly research and outputs that engage with collaborators and audiences not only including our colleagues, but also students, academics in different fields including the hard sciences, professionals and the broader public. Collaboration and scholarly interaction, particularly with better-funded and more technically advanced disciplines, is essential to digital humanities and perhaps even more so to digital classics. The international perspectives on these issues are especially valuable in an increasingly connected, institutionally and administratively diverse world.{\textless}br{\textgreater}{\textless}br{\textgreater}This book addresses the broad range of issues scholars and practitioners face in engaging with students, professionals and the public, in accessible and valuable chapters from authors of many backgrounds and areas of expertise, including language and linguistics, history, archaeology and architecture. This collection will be of interest to teachers, scientists, cultural heritage professionals, linguists and enthusiasts of history and antiquity.},
	language = {en},
	urldate = {2020-02-04},
	booktitle = {Digital {Classics} {Outside} the {Echo}-{Chamber}},
	publisher = {Ubiquity Press},
	author = {Mahony, Simon},
	editor = {Romanello, Matteo and Bodard, Gabriel},
	month = apr,
	year = {2016},
	keywords = {reference},
	pages = {33--50},
}

Edited by organizers of “Digital Classicist” seminars in London and Berlin, this volume addresses the impact of computational approaches to the study of antiquity on audiences other than the scholars who conventionally publish it. In addition to colleagues in classics and digital humanities, the eleven chapters herein concern and are addressed to students, heritage professionals and “citizen scientists”.\textlessbr\textgreater\textlessbr\textgreaterEach chapter is a scholarly contribution, presenting research questions in the classics, digital humanities or, in many cases, both. They are all also examples of work within one of the most important areas of academia today: scholarly research and outputs that engage with collaborators and audiences not only including our colleagues, but also students, academics in different fields including the hard sciences, professionals and the broader public. Collaboration and scholarly interaction, particularly with better-funded and more technically advanced disciplines, is essential to digital humanities and perhaps even more so to digital classics. The international perspectives on these issues are especially valuable in an increasingly connected, institutionally and administratively diverse world.\textlessbr\textgreater\textlessbr\textgreaterThis book addresses the broad range of issues scholars and practitioners face in engaging with students, professionals and the public, in accessible and valuable chapters from authors of many backgrounds and areas of expertise, including language and linguistics, history, archaeology and architecture. This collection will be of interest to teachers, scientists, cultural heritage professionals, linguists and enthusiasts of history and antiquity.

@inproceedings{nivre_universal_2016,
	title = {Universal dependencies v1: {A} multilingual treebank collection},
	url = {https://www.aclweb.org/anthology/L16-1262.pdf},
	author = {Nivre, Joakim and De Marneffe, Marie-Catherine and Ginter, Filip and Goldberg, Yoav and Hajic, Jan and Manning, Christopher D and McDonald, Ryan and Petrov, Slav and Pyysalo, Sampo and Silveira, Natalia},
	year = {2016},
	pages = {1659--1666},
}

@article{kirby_computational_2016,
	title = {A {Computational} {Method} for {Comparative} {Greek} and {Latin} {Prosimetrics}},
	language = {en},
	author = {Kirby, Joseph Tyler},
	year = {2016},
}

@article{stover_reassessing_2016,
	title = {{REASSESSING} {THE} {APULEIAN} {CORPUS}: {A} {COMPUTATIONAL} {APPROACH} {TO} {AUTHENTICITY}},
	volume = {66},
	issn = {0009-8388, 1471-6844},
	shorttitle = {{REASSESSING} {THE} {APULEIAN} {CORPUS}},
	url = {https://www.cambridge.org/core/journals/classical-quarterly/article/abs/reassessing-the-apuleian-corpus-a-computational-approach-to-authenticity/6BFD216C8AC552673B16BBA0FA40FF54},
	doi = {10.1017/S0009838816000768},
	abstract = {The renaissance of Apuleian studies of the past few decades shows no signs of abating.1 The summer of 2014 may well be the highest watermark yet recorded in the tide of interest in Apuleius: June and July alone saw the release of two monographs, one each from Oxford University Press and Cambridge University Press, and one edited conference volume, from Routledge.2 The clearest sign that the sophist of Madauros has come into his own is his admission into the exclusive club of the Oxford Classical Texts: the first volume of his complete works containing the Metamorphoses edited by Maaike Zimmerman came out in 2012. One of the most salutary effects of this renewed interest has been the reappraisal of the ‘whole Apuleius’: Apuleius has more to offer than just the Metamorphoses, and recent scholarship on the rhetorica and the philosophica have shown not only how these opera minora can help us understand the opus maius, but also how they are important and interesting documents in their own right.3},
	language = {en},
	number = {2},
	urldate = {2023-08-26},
	journal = {The Classical Quarterly},
	publisher = {Cambridge University Press},
	author = {Stover, Justin and Kestemont, Mike},
	month = dec,
	year = {2016},
	pages = {645--672},
}

The renaissance of Apuleian studies of the past few decades shows no signs of abating.1 The summer of 2014 may well be the highest watermark yet recorded in the tide of interest in Apuleius: June and July alone saw the release of two monographs, one each from Oxford University Press and Cambridge University Press, and one edited conference volume, from Routledge.2 The clearest sign that the sophist of Madauros has come into his own is his admission into the exclusive club of the Oxford Classical Texts: the first volume of his complete works containing the Metamorphoses edited by Maaike Zimmerman came out in 2012. One of the most salutary effects of this renewed interest has been the reappraisal of the ‘whole Apuleius’: Apuleius has more to offer than just the Metamorphoses, and recent scholarship on the rhetorica and the philosophica have shown not only how these opera minora can help us understand the opus maius, but also how they are important and interesting documents in their own right.3

@inproceedings{elsner_automatic_2016,
	address = {Berlin, Germany},
	title = {Automatic discovery of {Latin} syntactic changes},
	url = {http://aclweb.org/anthology/W16-2120},
	doi = {10.18653/v1/W16-2120},
	abstract = {Syntactic change tends to affect constructions, but treebanks annotate lower-level structure: PCFG rules or dependency arcs. This paper extends prior work in native language identification, using Tree Substitution Grammars to discover constructions which can be tested for historical variability. In a case study comparing Classical and Medieval Latin, the system discovers several constructions corresponding to known historical differences, and learns to distinguish the two varieties with high accuracy. Applied to an intermediate text (the Vulgate Bible), it indicates which changes between the eras were already occurring at this earlier stage.},
	language = {en},
	urldate = {2023-08-26},
	booktitle = {Proceedings of the 10th {SIGHUM} {Workshop} on {Language} {Technology} for           {Cultural} {Heritage}, {Social} {Sciences}, and {Humanities}},
	publisher = {Association for Computational Linguistics},
	author = {Elsner, Micha and Lane, Emily},
	year = {2016},
	pages = {156--164},
}

Syntactic change tends to affect constructions, but treebanks annotate lower-level structure: PCFG rules or dependency arcs. This paper extends prior work in native language identification, using Tree Substitution Grammars to discover constructions which can be tested for historical variability. In a case study comparing Classical and Medieval Latin, the system discovers several constructions corresponding to known historical differences, and learns to distinguish the two varieties with high accuracy. Applied to an intermediate text (the Vulgate Bible), it indicates which changes between the eras were already occurring at this earlier stage.

@article{field_automated_2016,
	title = {An {Automated} {Approach} to {Syntax}-based {Analysis} of {Classical} {Latin}},
	copyright = {Copyright (c) 2016 Anjalie Field},
	issn = {2364-7957},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/32315},
	doi = {10.11588/dco.2016.0.32315},
	abstract = {The goal of this study is to present an automated method for analyzing the style of Latin authors. Many of the common automated methods in stylistic analysis are based on lexical measures, which do not work well with Latin because of the language’s high degree of inflection and free word order. In contrast, this study focuses on analysis at a syntax level by examining two constructions, the ablative absolute and the cum clause. These constructions are often interchangeable, which suggests an author’s choice of construction is typically more stylistic than functional. We first identified these constructions in hand-annotated texts. Next we developed a method for identifying the constructions in unannotated texts, using probabilistic morphological tagging. Our methods identified constructions with enough accuracy to distinguish among different genres and different authors. In particular, we were able to determine which book of Caesar’s Commentarii de Bello Gallico was not written by Caesar. Furthermore, the usage of ablative absolutes and cum clauses observed in this study is consistent with the usage scholars have observed when analyzing these texts by hand. The proposed methods for an automatic syntax-based analysis are shown to be valuable for the study of classical literature.},
	language = {en},
	urldate = {2023-08-26},
	journal = {Digital Classics Online},
	author = {Field, Anjalie},
	month = dec,
	year = {2016},
	pages = {57--78},
}

The goal of this study is to present an automated method for analyzing the style of Latin authors. Many of the common automated methods in stylistic analysis are based on lexical measures, which do not work well with Latin because of the language’s high degree of inflection and free word order. In contrast, this study focuses on analysis at a syntax level by examining two constructions, the ablative absolute and the cum clause. These constructions are often interchangeable, which suggests an author’s choice of construction is typically more stylistic than functional. We first identified these constructions in hand-annotated texts. Next we developed a method for identifying the constructions in unannotated texts, using probabilistic morphological tagging. Our methods identified constructions with enough accuracy to distinguish among different genres and different authors. In particular, we were able to determine which book of Caesar’s Commentarii de Bello Gallico was not written by Caesar. Furthermore, the usage of ablative absolutes and cum clauses observed in this study is consistent with the usage scholars have observed when analyzing these texts by hand. The proposed methods for an automatic syntax-based analysis are shown to be valuable for the study of classical literature.

@article{kestemont_authenticating_2016,
	title = {Authenticating the writings of {Julius} {Caesar}},
	volume = {63},
	issn = {09574174},
	url = {https://linkinghub.elsevier.com/retrieve/pii/S0957417416303116},
	doi = {10.1016/j.eswa.2016.06.029},
	language = {en},
	urldate = {2023-07-23},
	journal = {Expert Systems with Applications},
	author = {Kestemont, Mike and Stover, Justin and Koppel, Moshe and Karsdorp, Folgert and Daelemans, Walter},
	month = nov,
	year = {2016},
	pages = {86--96},
}

@book{speich_algorithms_2016,
	address = {Düsseldorf},
	title = {Algorithms and {Aristotle}},
	isbn = {978-3-945627-13-6},
	abstract = {A glossary of key terms on digitalization and its effects on the labour market and education can be found in alphabetical order throughout the book},
	publisher = {Vodafone Foundation Germany},
	editor = {Speich, M.},
	year = {2016},
}

A glossary of key terms on digitalization and its effects on the labour market and education can be found in alphabetical order throughout the book

@article{stover_authorship_2016,
	title = {The {Authorship} of the "{Historia} {Augusta}": {Two} {New} {Computational} {Studies}},
	volume = {59},
	issn = {0076-0730},
	shorttitle = {The {Authorship} of the "{Historia} {Augusta}"},
	url = {https://www.jstor.org/stable/44254158},
	abstract = {The case of the Historia Augusta, a collection of imperial biographies from Hadrian to Cams supposedly written by six different authors, provided the impetus for the introduction of computational methods into the Echtheitskritik of ancient authors in 1979. After a flurry of studies in the 1990s, interest waned, particularly because most of those studies seemed to support conclusions incompatible with the scholarly consensus on the question. In the paper, we approach this question with the new tool of authorship verification – one of the most promising approaches in forensic stylometry today – as well as the established method of principal components analysis to demonstrate that there is no simple alternative between single and multiple authorship, and that the results of a computational analysis are in fact compatible with the results obtained from historical, literary, and philological analysis.},
	number = {2},
	urldate = {2023-07-23},
	journal = {Bulletin of the Institute of Classical Studies},
	publisher = {Wiley},
	author = {Stover, Justin A. and Kestemont, Mike},
	year = {2016},
	pages = {140--157},
}

The case of the Historia Augusta, a collection of imperial biographies from Hadrian to Cams supposedly written by six different authors, provided the impetus for the introduction of computational methods into the Echtheitskritik of ancient authors in 1979. After a flurry of studies in the 1990s, interest waned, particularly because most of those studies seemed to support conclusions incompatible with the scholarly consensus on the question. In the paper, we approach this question with the new tool of authorship verification – one of the most promising approaches in forensic stylometry today – as well as the established method of principal components analysis to demonstrate that there is no simple alternative between single and multiple authorship, and that the results of a computational analysis are in fact compatible with the results obtained from historical, literary, and philological analysis.

@inproceedings{straka_udpipe_2016,
	title = {{UDPipe}: {Trainable} {Pipeline} for {Processing} {CoNLL}-{U} {Files} {Performing} {Tokenization}, {Morphological} {Analysis}, {POS} {Tagging} and {Parsing}},
	booktitle = {{LREC}},
	author = {Straka, Milan and Hajic, Jan and Straková, Jana},
	year = {2016},
	pages = {4290--4297},
}

@misc{blackwell_canonical_2015,
	title = {The {Canonical} {Text} {Services} {URN} {Specification}, {Version} 2.0.rc.1},
	shorttitle = {{CITE} / {URN}},
	url = {http://cite-architecture.github.io/ctsurn_spec/},
	author = {Blackwell, Christopher and Smith, Neel},
	year = {2015},
	keywords = {reference},
}

@book{ridsdale_strategies_2015,
	address = {Halifax},
	title = {Strategies and {Best} {Practices} for {Data} {Literacy} {Education} {Knowledge} {Synthesis} {Report}},
	url = {https://dalspaceb.library.dal.ca/server/api/core/bitstreams/7c0db868-79ac-4c3c-9ad0-f2a93cf2ccc6/content},
	doi = {10.13140/RG.2.1.1922.5044},
	abstract = {Key Messages BACKGROUND: We begin with a definition, synthesized from existing literature and refined based on expert input: Data literacy is the ability to collect, manage, evaluate, and apply data, in a critical manner. It is an essential ability required in the global knowledge-­‐‑ based economy; the manipulation of data occurs in daily processes across all sectors and disciplines. An understanding of how decisions are informed by data, and how to collect, manage, evaluate, and apply this data in support of evidence-­‐‑ based decision-­‐‑ making, will benefit Canadian citizens, and will increasingly be required in knowledge economy jobs. Data literacy education is currently inconsistent across the public, private, and academic sectors, and data literacy training has not been approached systematically or formally at Canada'ʹs post-­‐‑ secondary institutions. There are also per-­‐‑ sector capability gaps, which makes it difficult to set realistic expectations of data-­‐‑ based skills.},
	publisher = {Dalhousie University},
	author = {Ridsdale, Chantel and Rothwell, James and Smit, Mike and Bliemel, Michael and Irvine, Dean and Kelley, Dan and Matwin, Stan and Wuetherick, Brad and Ali-Hassan, Hossam},
	month = jan,
	year = {2015},
}

Key Messages BACKGROUND: We begin with a definition, synthesized from existing literature and refined based on expert input: Data literacy is the ability to collect, manage, evaluate, and apply data, in a critical manner. It is an essential ability required in the global knowledge-­‐‑ based economy; the manipulation of data occurs in daily processes across all sectors and disciplines. An understanding of how decisions are informed by data, and how to collect, manage, evaluate, and apply this data in support of evidence-­‐‑ based decision-­‐‑ making, will benefit Canadian citizens, and will increasingly be required in knowledge economy jobs. Data literacy education is currently inconsistent across the public, private, and academic sectors, and data literacy training has not been approached systematically or formally at Canada'ʹs post-­‐‑ secondary institutions. There are also per-­‐‑ sector capability gaps, which makes it difficult to set realistic expectations of data-­‐‑ based skills.

@misc{scheidel_orbis_2015,
	address = {Rochester, NY},
	type = {{SSRN} {Scholarly} {Paper}},
	title = {Orbis: {The} {Stanford} {Geospatial} {Network} {Model} of the {Roman} {World}},
	shorttitle = {Orbis},
	url = {https://papers.ssrn.com/abstract=2609654},
	doi = {10.2139/ssrn.2609654},
	abstract = {ORBIS allows us to express ancient Roman communication costs in terms of both time and expense. By simulating movement along the principal routes of the Roman road network, the main navigable rivers, and hundreds of sea routes in the Mediterranean, Black Sea and coastal Atlantic, this interactive model reconstructs the duration and financial cost of travel in antiquity. Taking account of seasonal variation and accommodating a wide range of modes and means of transport, ORBIS reveals the true shape of the Roman world and provides a unique resource for our understanding of premodern history.},
	language = {en},
	urldate = {2025-06-20},
	publisher = {Social Science Research Network},
	author = {Scheidel, Walter},
	month = may,
	year = {2015},
	keywords = {Communication, Geospatial modeling, Networks, Transport, Travel},
}

ORBIS allows us to express ancient Roman communication costs in terms of both time and expense. By simulating movement along the principal routes of the Roman road network, the main navigable rivers, and hundreds of sea routes in the Mediterranean, Black Sea and coastal Atlantic, this interactive model reconstructs the duration and financial cost of travel in antiquity. Taking account of seasonal variation and accommodating a wide range of modes and means of transport, ORBIS reveals the true shape of the Roman world and provides a unique resource for our understanding of premodern history.

@misc{dunning_rethinking_2015,
	title = {Rethinking the publication of premodern sources: {Petrus} {Plaoul} on the {Sentences} – {RIDE}},
	shorttitle = {Rethinking the publication of premodern sources},
	url = {https://ride.i-d-e.de/issues/issue-3/petrus_plaoul/},
	language = {en-US},
	urldate = {2025-06-19},
	author = {Dunning, Andrew},
	year = {2015},
}

@article{schubert_close_2015,
	title = {Close {Reading} und {Distant} {Reading}. {Methoden} der {Altertumswissenschaften} in der {Gegenwart}},
	copyright = {Copyright (c) 2015 Charlotte Schubert},
	issn = {2364-7957},
	url = {https://journals.ub.uni-heidelberg.de/index.php/dco/article/view/20483},
	doi = {10.11588/dco.2015.1.20483},
	abstract = {\_},
	language = {de},
	urldate = {2025-02-02},
	journal = {Digital Classics Online},
	author = {Schubert, Charlotte},
	month = may,
	year = {2015},
	keywords = {Visualisierung},
	pages = {1--6},
}

_

@inproceedings{broux_developing_2015,
	address = {Cham},
	title = {Developing {Onomastic} {Gazetteers} and {Prosopographies} for the {Ancient} {World} {Through} {Named} {Entity} {Recognition} and {Graph} {Visualization}: {Some} {Examples} from {Trismegistos} {People}},
	isbn = {978-3-319-15168-7},
	shorttitle = {Developing {Onomastic} {Gazetteers} and {Prosopographies} for the {Ancient} {World} {Through} {Named} {Entity} {Recognition} and {Graph} {Visualization}},
	doi = {10.1007/978-3-319-15168-7_38},
	abstract = {Developing prosopographies or onomastic lists in a non-digital environment used to be a painstaking and time-consuming exercise, involving manual labour by teams of researchers, often taking decades. For some scholarly disciplines from the ancient world this is still true, especially those studying non-alphabetical writing systems that lack a uniform transcription system, e.g. Demotic. But for many others, such as Greek and Latin, digital full text corpora in Unicode are now available, often even freely accessible. In this paper we illustrate, on the basis of Trismegistos, how data collection through Named Entity Recognition and visualization through Social Network Analysis have huge potential to speed up the creation of onomastic lists and the development of prosopographies.},
	language = {en},
	booktitle = {Social {Informatics}},
	publisher = {Springer International Publishing},
	author = {Broux, Yanne and Depauw, Mark},
	editor = {Aiello, Luca Maria and McFarland, Daniel},
	year = {2015},
	pages = {304--313},
}

Developing prosopographies or onomastic lists in a non-digital environment used to be a painstaking and time-consuming exercise, involving manual labour by teams of researchers, often taking decades. For some scholarly disciplines from the ancient world this is still true, especially those studying non-alphabetical writing systems that lack a uniform transcription system, e.g. Demotic. But for many others, such as Greek and Latin, digital full text corpora in Unicode are now available, often even freely accessible. In this paper we illustrate, on the basis of Trismegistos, how data collection through Named Entity Recognition and visualization through Social Network Analysis have huge potential to speed up the creation of onomastic lists and the development of prosopographies.

@article{eckhoff_linguistics_2015,
	title = {Linguistics vs. digital editions: {The} {Tromsø} {Old} {Russian} and {OCS} {Treebank}},
	url = {https://munin.uit.no/bitstream/handle/10037/22366/article.pdf?sequence=2&isAllowed=y},
	journal = {Scripta \& e-Scripta},
	author = {Eckhoff, H. M. and Bredicevskis, A.},
	year = {2015},
	pages = {9--25},
}

@article{fusi_multilanguage_2015,
	title = {A {Multilanguage}, {Modular} {Framework} for {Metrical} {Analysis} : {It} {Patterns} and {Theorical} {Issues}},
	volume = {199},
	doi = {doi:10.3917/lang.199.0041.},
	journal = {Langages},
	author = {Fusi, DanieleFusi},
	year = {2015},
	pages = {41--66},
}

@article{de_felice_classes_2015,
	title = {{CLaSSES} : {A} {New} {Digital} {Resource} for {Latin} {Epigraphy}},
	volume = {1},
	doi = {doi:10.1418/90427},
	number = {1},
	journal = {Italian journal of computational linguistics},
	author = {De Felice, Irene and Donati, Margherita and Giovanna, Marotta},
	year = {2015},
	pages = {125--136},
}

@article{mcgillivray_computational_2015,
	title = {Computational {Valency} {Lexica} for {Latin} and {Greek} in {Use} : {A} {Case} {Study} of {Syntactic} {Ambiguity}},
	volume = {14},
	url = {doi:10.1515/joll-2015-0005},
	number = {1},
	journal = {Journal of Latin Linguistics},
	author = {McGillivray, Barbara and Vatri, Alessandro},
	year = {2015},
	pages = {101--126},
}

@inproceedings{bjerva_word_2015,
	address = {Beijing, China},
	title = {Word {Embeddings} {Pointing} the {Way} for {Late} {Antiquity}},
	url = {https://www.aclweb.org/anthology/W15-3708},
	doi = {10.18653/v1/W15-3708},
	urldate = {2021-04-24},
	booktitle = {Proceedings of the 9th {SIGHUM} {Workshop} on {Language} {Technology} for {Cultural} {Heritage}, {Social} {Sciences}, and {Humanities} ({LaTeCH})},
	publisher = {Association for Computational Linguistics},
	author = {Bjerva, Johannes and Praet, Raf},
	month = jul,
	year = {2015},
	pages = {53--57},
}

@book{richards_software_2015,
	title = {Software {Architecture} {Patterns}},
	url = {http://103.62.146.201:8081/xmlui/handle/1/5665},
	abstract = {It’s all too common for developers to start coding an application 
without a formal architecture in place. Without a clear and well- 
defined architecture, most developers and architects will resort to},
	language = {en},
	urldate = {2023-08-31},
	author = {Richards, Mark},
	year = {2015},
	note = {Accepted: 2018-01-19T11:08:40Z},
}

It’s all too common for developers to start coding an application without a formal architecture in place. Without a clear and well- defined architecture, most developers and architects will resort to

@article{simon_linking_2015,
	title = {Linking early geospatial documents, one place at a time: annotation of geographic documents with {Recogito}},
	volume = {10},
	issn = {1790-3769},
	shorttitle = {Linking early geospatial documents, one place at a time},
	url = {http://www.e-perimetron.org/Vol_10_2/Simon_et_al.pdf},
	abstract = {Recogito is an open source tool for the semi-automatic annotation of place references in maps and texts. It was developed as part of the Pelagios 3 research project, which aims to build up a comprehensive directory of places referred to in early maps and geographic writing predating the year 1492. Pelagios 3 focuses specifically on sources from the Classical Latin, Greek and Byzantine periods; on Mappae Mundi and narrative texts from the European Medieval period; on Late Medieval Portolans; and on maps and texts from the early Islamic and early Chinese traditions. Since the start of the project in September 2013, the team has harvested more than 120,000 toponyms, manually verifying almost 60,000 of them. Furthermore, the team held two public annotation workshops supported through the Open Humanities Awards 2014. In these workshops, a mixed audience of students and academics of different backgrounds used Recogito to add several thousand contributions on each workshop day.

A number of benefits arise out of this work: on the one hand, the digital identification of places – and the names used for them – makes the documents' contents amenable to information retrieval technology, i.e. documents become more easily search- and discoverable to users than through conventional metadata-based search alone. On the other hand, the documents are opened up to new forms of re-use. For example, it becomes possible to “map” and compare the narrative of texts, and the contents of maps with modern day tools like Web maps and GIS; or to analyze and contrast documents’ geographic properties, toponymy and spatial relationships. Seen in a wider context, we argue that initiatives such as ours contribute to the growing ecosystem of the “Graph of Humanities Data” that is gathering pace in the Digital Humanities (linking data about people, places, events, canonical references, etc.), which has the potential to open up new avenues for computational and quantitative research in a variety of fields including History, Geography, Archaeology, Classics, Genealogy and Modern Languages.},
	language = {en},
	number = {2},
	urldate = {2023-08-26},
	journal = {e-Perimetron},
	author = {Simon, Rainer and Barker, Elton and Isaksen, Leif and de Soto Cañamares, Pau},
	year = {2015},
	note = {Number: 2},
	pages = {49--59},
}

Recogito is an open source tool for the semi-automatic annotation of place references in maps and texts. It was developed as part of the Pelagios 3 research project, which aims to build up a comprehensive directory of places referred to in early maps and geographic writing predating the year 1492. Pelagios 3 focuses specifically on sources from the Classical Latin, Greek and Byzantine periods; on Mappae Mundi and narrative texts from the European Medieval period; on Late Medieval Portolans; and on maps and texts from the early Islamic and early Chinese traditions. Since the start of the project in September 2013, the team has harvested more than 120,000 toponyms, manually verifying almost 60,000 of them. Furthermore, the team held two public annotation workshops supported through the Open Humanities Awards 2014. In these workshops, a mixed audience of students and academics of different backgrounds used Recogito to add several thousand contributions on each workshop day. A number of benefits arise out of this work: on the one hand, the digital identification of places – and the names used for them – makes the documents' contents amenable to information retrieval technology, i.e. documents become more easily search- and discoverable to users than through conventional metadata-based search alone. On the other hand, the documents are opened up to new forms of re-use. For example, it becomes possible to “map” and compare the narrative of texts, and the contents of maps with modern day tools like Web maps and GIS; or to analyze and contrast documents’ geographic properties, toponymy and spatial relationships. Seen in a wider context, we argue that initiatives such as ours contribute to the growing ecosystem of the “Graph of Humanities Data” that is gathering pace in the Digital Humanities (linking data about people, places, events, canonical references, etc.), which has the potential to open up new avenues for computational and quantitative research in a variety of fields including History, Geography, Archaeology, Classics, Genealogy and Modern Languages.

@article{revellio_classics_2015,
	title = {Classics and the {Digital} {Age} {Advantages} and limitations of digital text analysis in classical philology},
	issn = {2364-5342},
	url = {https://kops.uni-konstanz.de/server/api/core/bitstreams/620defc4-effd-4224-bfb6-782e20748e01/content},
	abstract = {Die Klassische Philologie nahm computergestützte Methoden der Textanalyse früh als Chance wahr. Um einen Einblick in die jüngsten Entwicklungen der digitalen Textanalyse im Bereich der Latinistik zu geben wird eine Auswahl bestehender Textdatenbanken wie gängiger Analysetools vorgestellt, wobei insbesondere auf das Phänomen der Intertextualität als Untersuchungsfeld fokussiert wird. Zudem werden unmittelbar
verknüpfte Themen wie die Digitalisierung und langfristige Erhaltung antiker Texte, der Status unterschiedlicher Text-Surrogate sowie die Notwendigkeit fremdsprachlicher Kenntnisse diskutiert.},
	urldate = {2023-08-26},
	author = {Revellio, Marie},
	year = {2015},
	pages = {1--16},
}

Die Klassische Philologie nahm computergestützte Methoden der Textanalyse früh als Chance wahr. Um einen Einblick in die jüngsten Entwicklungen der digitalen Textanalyse im Bereich der Latinistik zu geben wird eine Auswahl bestehender Textdatenbanken wie gängiger Analysetools vorgestellt, wobei insbesondere auf das Phänomen der Intertextualität als Untersuchungsfeld fokussiert wird. Zudem werden unmittelbar verknüpfte Themen wie die Digitalisierung und langfristige Erhaltung antiker Texte, der Status unterschiedlicher Text-Surrogate sowie die Notwendigkeit fremdsprachlicher Kenntnisse diskutiert.

@book{blackwell_canonical_2015,
	title = {The {Canonical} {Text} {Services} {URN} {Specification}, {Version} 2.0.{Rc}.1 [{CITE} / {URN}]},
	author = {Blackwell, Christopher and Smith, Neel},
	year = {2015},
}

@inproceedings{perkins_building_2014,
	title = {Building {Bridges} to the {Future} of a {Distributed} {Network}: {From} {DiRT} {Categories} to {TaDiRAH}, a {Methods} {Taxonomy} for {Digital} {Humanities}},
	copyright = {https://creativecommons.org/licenses/by/4.0/},
	shorttitle = {Building {Bridges} to the {Future} of a {Distributed} {Network}},
	url = {https://dcpapers.dublincore.org/article/952136591},
	doi = {10.23106/dcmi.952136591},
	abstract = {Efforts to establish centralized hubs of information relevant to digital humanities (DH) have proven unsustainable over the long term. Comprehensive hubs are currently being re-designed with a smaller scope and focused curation. However, this smaller scope comes with the risk of decontextualization -- a digital humanities project is best understood through the intersection of its subject matter, methodologies and applications, not all of which are captured by any single site. This poster will trace the development and application of 'TaDiRAH' a shared taxonomy of digital humanities research activities and objects, created for the purpose of bridging the divide between related digital humanities hubs.},
	language = {en},
	urldate = {2026-02-05},
	booktitle = {Proceedings of the {International} {Conference} on {Dublin} {Core} and {Metadata} {Applications}},
	publisher = {Dublin Core Metadata Initiative},
	author = {Perkins, Jody and Dombrowski, Quinn and Borek, Luise and Schöch, Christof},
	month = oct,
	year = {2014},
}

Efforts to establish centralized hubs of information relevant to digital humanities (DH) have proven unsustainable over the long term. Comprehensive hubs are currently being re-designed with a smaller scope and focused curation. However, this smaller scope comes with the risk of decontextualization – a digital humanities project is best understood through the intersection of its subject matter, methodologies and applications, not all of which are captured by any single site. This poster will trace the development and application of 'TaDiRAH' a shared taxonomy of digital humanities research activities and objects, created for the purpose of bridging the divide between related digital humanities hubs.

@inproceedings{bizzoni_making_2014,
	title = {The {Making} of {Ancient} {Greek} {WordNet}},
	volume = {2014},
	url = {http://www.lrec-conf.org/proceedings/lrec2014/pdf/1071_Paper.pdf},
	author = {Bizzoni, Yuri and Boschetti, Federico and Diakoff, Harry and Del Gratta, Riccardo and Monachini, Monica and Crane, Gregory R},
	year = {2014},
	pages = {1140--1147},
}

@inproceedings{broux_developing_2014,
	title = {Developing onomastic gazetteers and prosopographies for the ancient world through named entity recognition and graph visualization: {Some} examples from trismegistos people},
	url = {https://lirias.kuleuven.be/retrieve/285982},
	booktitle = {Social {Informatics}. {SocInfo} 2014},
	publisher = {Springer},
	author = {Broux, Yanne and Depauw, Mark},
	editor = {Aiello, Luca Maria and McFarland, Daniel},
	year = {2014},
	pages = {304--313},
}

@article{coffee_tesserae_2014,
	title = {Tesserae: {A} {Search} {Engine} for {Allusion}},
	shorttitle = {Tesserae},
	url = {https://hcommons.org/deposits/item/hc:12221/},
	abstract = {The Tesserae Project is an interdisciplinary research effort employing computational methods to detect and analyze literary allusion (a form of text reuse) currently focusing on Latin and ancient Greek. The Project seeks funding to create a fully-functional, publicly available tool to detect similar phrases in two texts at rates that approach those of literary commentators. To this end, funding will support adding sensitivity to word meaning, phrase context, and sound similarity. Detection rate improvements will be measured against a set of 3000 parallel phrases previously graded for literary significance. A revised website will inform researchers of research results and new functions of the tool. The project team will give presentations and produce publications explaining the function, results, and theoretical consequences of the fully operational tool. This work is preliminary to an out-year Implementation Phase that will see the addition of English, French, Italian, and Spanish.},
	language = {en-US},
	urldate = {2023-08-26},
	author = {Coffee, Neil and Koenig, Jean-Pierre},
	month = may,
	year = {2014},
}

The Tesserae Project is an interdisciplinary research effort employing computational methods to detect and analyze literary allusion (a form of text reuse) currently focusing on Latin and ancient Greek. The Project seeks funding to create a fully-functional, publicly available tool to detect similar phrases in two texts at rates that approach those of literary commentators. To this end, funding will support adding sensitivity to word meaning, phrase context, and sound similarity. Detection rate improvements will be measured against a set of 3000 parallel phrases previously graded for literary significance. A revised website will inform researchers of research results and new functions of the tool. The project team will give presentations and produce publications explaining the function, results, and theoretical consequences of the fully operational tool. This work is preliminary to an out-year Implementation Phase that will see the addition of English, French, Italian, and Spanish.

@inproceedings{passarotti_syntax_2014,
	address = {Gothenburg, Sweden},
	title = {From {Syntax} to {Semantics}. {First} {Steps} {Towards} {Tectogrammatical} {Annotation} of {Latin}},
	url = {http://aclweb.org/anthology/W14-0615},
	doi = {10.3115/v1/W14-0615},
	abstract = {Assuming that collaboration between theoretical and computational linguistics is essential in projects aimed at developing language resources like annotated corpora, this paper presents the first steps of the semantic annotation of the Index Thomisticus Treebank, a dependency-based treebank of Medieval Latin. The semantic layer of annotation of the treebank is detailed and the theoretical framework supporting the annotation style is explained and motivated.},
	language = {en},
	urldate = {2023-08-26},
	booktitle = {Proceedings of the 8th {Workshop} on {Language} {Technology} for {Cultural} {Heritage}, {Social} {Sciences}, and {Humanities} ({LaTeCH})},
	publisher = {Association for Computational Linguistics},
	author = {Passarotti, Marco},
	year = {2014},
	pages = {100--109},
}

Assuming that collaboration between theoretical and computational linguistics is essential in projects aimed at developing language resources like annotated corpora, this paper presents the first steps of the semantic annotation of the Index Thomisticus Treebank, a dependency-based treebank of Medieval Latin. The semantic layer of annotation of the treebank is detailed and the theoretical framework supporting the annotation style is explained and motivated.

@article{almas_linked_2014,
	title = {Linked {Data} in the {Perseus} {Digital} {Library}},
	volume = {7},
	url = {http://dlib.nyu.edu/awdl/isaw/isaw-papers/7/almas-babeu-krohn/},
	number = {3},
	journal = {ISAW Papers},
	author = {Almas, B. and Babeu, A. and Krohn, A.},
	year = {2014},
}

@article{koppel_determining_2014,
	title = {Determining if two documents are written by the same author},
	volume = {65},
	issn = {2330-1643},
	url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/asi.22954},
	doi = {10.1002/asi.22954},
	abstract = {Almost any conceivable authorship attribution problem can be reduced to one fundamental problem: whether a pair of (possibly short) documents were written by the same author. In this article, we offer an (almost) unsupervised method for solving this problem with surprisingly high accuracy. The main idea is to use repeated feature subsampling methods to determine if one document of the pair allows us to select the other from among a background set of “impostors” in a sufficiently robust manner.},
	language = {en},
	number = {1},
	urldate = {2023-07-23},
	journal = {Journal of the Association for Information Science and Technology},
	author = {Koppel, Moshe and Winter, Yaron},
	year = {2014},
	note = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/asi.22954},
	keywords = {automatic classification, machine learning, natural language processing},
	pages = {178--187},
}

Almost any conceivable authorship attribution problem can be reduced to one fundamental problem: whether a pair of (possibly short) documents were written by the same author. In this article, we offer an (almost) unsupervised method for solving this problem with surprisingly high accuracy. The main idea is to use repeated feature subsampling methods to determine if one document of the pair allows us to select the other from among a background set of “impostors” in a sufficiently robust manner.

@incollection{tiepmar_new_2014,
	title = {A new implementation for canonical text services [{CTS}]},
	isbn = {978-1-63266-401-3},
	booktitle = {Proceedings of the 8th {Workshop} on {Language} {Technology} for {Cultural} {Heritage}, {Social} {Sciences}, and {Humanities} ({LaTeCH})},
	author = {Tiepmar, J. and Teichmann, C. and Heyer, G. and Berti, M. and Crane, G.},
	year = {2014},
	pages = {1--8},
}

@incollection{jannidis_theorien_2013,
	address = {Stuttgart},
	title = {Theorien und {Methoden} der {Literaturwissenschaft}},
	isbn = {978-3-476-01271-5},
	url = {https://doi.org/10.1007/978-3-476-01271-5_15},
	doi = {10.1007/978-3-476-01271-5_15},
	abstract = {Im literaturwissenschaftlichen Umgang mit Literatur sind mindestens sechs Ziele institutionell etabliert:1.Sichern der Textgrundlage, Herstellen verlässlicher Texte: Theorie und Methoden der Edition und Textkritik (vgl. II.1.1)2.Analysieren und Deuten der Texte: Methoden der Textanalyse und Interpretationstheorien3.Rekonstruktion der geschichtlichen Entwicklung von Literatur: Theorien und Methoden der Literaturgeschichtsschreibung (vgl. II.4)4.Analyse von Handlungen im Umgang mit Literatur: soziologische und psychologische Theorien und Methoden5.Erklärungen für das Phänomen ›Literatur‹: Evolutionsbiologische und anthropologische Theorien6.Wertung von Literatur (vgl. II.3)},
	language = {de},
	urldate = {2025-12-29},
	booktitle = {Handbuch {Literaturwissenschaft}: {Band} 1: {Gegenstände} und {Grundbegriffe} / {Band} 2: {Methoden} und {Theorien} / {Band} 3: {Institutionen} und {Praxisfelder}},
	publisher = {J.B. Metzler},
	author = {Jannidis, Fotis and Köppe, Tilmann and Winko, Simone},
	editor = {Anz, Thomas},
	year = {2013},
	pages = {797--883},
}

Im literaturwissenschaftlichen Umgang mit Literatur sind mindestens sechs Ziele institutionell etabliert:1.Sichern der Textgrundlage, Herstellen verlässlicher Texte: Theorie und Methoden der Edition und Textkritik (vgl. II.1.1)2.Analysieren und Deuten der Texte: Methoden der Textanalyse und Interpretationstheorien3.Rekonstruktion der geschichtlichen Entwicklung von Literatur: Theorien und Methoden der Literaturgeschichtsschreibung (vgl. II.4)4.Analyse von Handlungen im Umgang mit Literatur: soziologische und psychologische Theorien und Methoden5.Erklärungen für das Phänomen ›Literatur‹: Evolutionsbiologische und anthropologische Theorien6.Wertung von Literatur (vgl. II.3)

@book{moretti_distant_2013,
	title = {Distant {Reading}},
	isbn = {978-1-78168-333-0},
	abstract = {WINNER OF THE NATIONAL BOOK CRITICS CIRCLE AWARD How does a literary historian end up thinking in terms of z-scores, principal component analysis, and clustering coefficients? The essays in Distant Reading led to a new and often contested paradigm of literary analysis. In presenting them here Franco Moretti reconstructs his intellectual trajectory, the theoretical influences over his work, and explores the polemics that have often developed around his positions.From the evolutionary model of “Modern European Literature,” through the geo-cultural insights of “Conjectures of World Literature” and “Planet Hollywood,” to the quantitative findings of “Style, inc.” and the abstract patterns of “Network Theory, Plot Analysis,” the book follows two decades of conceptual development, organizing them around the metaphor of “distant reading,” that has come to define—well beyond the wildest expectations of its author—a growing field of unorthodox literary studies.},
	language = {en},
	publisher = {Verso Books},
	author = {Moretti, Franco},
	month = jun,
	year = {2013},
	note = {Google-Books-ID: Wo4110IdRLMC},
	keywords = {Literary Collections / Essays, Literary Criticism / Comparative Literature, Literary Criticism / Semiotics \& Theory, reference},
}

WINNER OF THE NATIONAL BOOK CRITICS CIRCLE AWARD How does a literary historian end up thinking in terms of z-scores, principal component analysis, and clustering coefficients? The essays in Distant Reading led to a new and often contested paradigm of literary analysis. In presenting them here Franco Moretti reconstructs his intellectual trajectory, the theoretical influences over his work, and explores the polemics that have often developed around his positions.From the evolutionary model of “Modern European Literature,” through the geo-cultural insights of “Conjectures of World Literature” and “Planet Hollywood,” to the quantitative findings of “Style, inc.” and the abstract patterns of “Network Theory, Plot Analysis,” the book follows two decades of conceptual development, organizing them around the metaphor of “distant reading,” that has come to define—well beyond the wildest expectations of its author—a growing field of unorthodox literary studies.

@book{mcgillivray_methods_2013,
	address = {Leiden; Boston},
	title = {Methods in {Latin} computational linguistics},
	url = {https://books.google.de/books?hl=de&lr=&id=33dfAgAAQBAJ},
	publisher = {Brill},
	author = {McGillivray, Barbara},
	year = {2013},
	keywords = {reference},
}

@book{moretti_distant_2013,
	address = {London; New York},
	title = {Distant {Reading}},
	isbn = {978-1-78168-333-0},
	url = {https://books.google.de/books?hl=en&lr=&id=Wo4110IdRLMC},
	abstract = {WINNER OF THE NATIONAL BOOK CRITICS CIRCLE AWARD How does a literary historian end up thinking in terms of z-scores, principal component analysis, and clustering coefficients? The essays in Distant Reading led to a new and often contested paradigm of literary analysis. In presenting them here Franco Moretti reconstructs his intellectual trajectory, the theoretical influences over his work, and explores the polemics that have often developed around his positions.From the evolutionary model of “Modern European Literature,” through the geo-cultural insights of “Conjectures of World Literature” and “Planet Hollywood,” to the quantitative findings of “Style, inc.” and the abstract patterns of “Network Theory, Plot Analysis,” the book follows two decades of conceptual development, organizing them around the metaphor of “distant reading,” that has come to define—well beyond the wildest expectations of its author—a growing field of unorthodox literary studies.},
	language = {en},
	publisher = {Verso Books},
	author = {Moretti, Franco},
	month = jun,
	year = {2013},
	note = {Google-Books-ID: Wo4110IdRLMC},
	keywords = {Literary Collections / Essays, Literary Criticism / Comparative Literature, Literary Criticism / Semiotics \& Theory, reference},
}

WINNER OF THE NATIONAL BOOK CRITICS CIRCLE AWARD How does a literary historian end up thinking in terms of z-scores, principal component analysis, and clustering coefficients? The essays in Distant Reading led to a new and often contested paradigm of literary analysis. In presenting them here Franco Moretti reconstructs his intellectual trajectory, the theoretical influences over his work, and explores the polemics that have often developed around his positions.From the evolutionary model of “Modern European Literature,” through the geo-cultural insights of “Conjectures of World Literature” and “Planet Hollywood,” to the quantitative findings of “Style, inc.” and the abstract patterns of “Network Theory, Plot Analysis,” the book follows two decades of conceptual development, organizing them around the metaphor of “distant reading,” that has come to define—well beyond the wildest expectations of its author—a growing field of unorthodox literary studies.

@article{eckhoff_mapping_2013,
	title = {Mapping out the {Source} {Domain} : {Evidence} from {Parallel} {Old} {Indo}-{European} {Data}},
	volume = {37},
	doi = {doi:10.1075/sl.37.2.03eck},
	number = {2},
	journal = {Studies in language},
	author = {Eckhoff, Hanne Martine and Thomason, Olga A. and Swart, Peter},
	year = {2013},
	pages = {302--355},
}

@article{mikolov_efficient_2013,
	title = {Efficient estimation of word representations in vector space},
	shorttitle = {{Word2Vec}},
	url = {https://arxiv.org/pdf/1301.3781.pdf},
	journal = {arXiv preprint arXiv:1301.3781},
	author = {Mikolov, Tomas and Chen, Kai and Corrado, Greg and Dean, Jeffrey},
	year = {2013},
	keywords = {reference},
	pages = {1--12},
}

@book{schmidt_pattern-oriented_2013,
	title = {Pattern-oriented software architecture, patterns for concurrent and networked objects},
	volume = {2},
	isbn = {1-118-72517-4},
	url = {https://www.researchgate.net/profile/Michael-Stal/publication/215835789_Pattern-Oriented_Software_Architecture_Patterns_for_Concurrent_and_Networked_Objects_Volume_2/links/53fd98440cf2dca8000356c5/Pattern-Oriented-Software-Architecture-Patterns-for-Concurrent-and-Networked-Objects-Volume-2.pdf},
	publisher = {John Wiley \& Sons},
	author = {Schmidt, Douglas C and Stal, Michael and Rohnert, Hans and Buschmann, Frank},
	year = {2013},
}

@article{brown_software_2013,
	title = {Software architecture for developers},
	url = {http://static.codingthearchitecture.com/sddconf2014-software-architecture-for-developers-extract.pdf},
	journal = {Coding the Architecture},
	author = {Brown, Simon},
	year = {2013},
}

@article{andrews_beyond_2013,
	title = {Beyond the tree of texts: {Building} an empirical model of scribal variation through graph analysis of texts and stemmata},
	volume = {28},
	issn = {0268-1145},
	shorttitle = {Beyond the tree of texts},
	url = {https://doi.org/10.1093/llc/fqt032},
	doi = {10.1093/llc/fqt032},
	abstract = {Stemmatology, or the reconstruction of the transmission history of texts, is a field that stands particularly to gain from digital methods. Many scholars already take stemmatic approaches that rely heavily on computational analysis of the collated text (e.g. Robinson and O’Hara 1996; Salemans 2000; Heikkilä 2005; Windram et al. 2008 among many others). Although there is great value in computationally assisted stemmatology, providing as it does a reproducible result and allowing access to the relevant methodological process in related fields such as evolutionary biology, computational stemmatics is not without its critics. The current state-of-the-art effectively forces scholars to choose between a preconceived judgment of the significance of textual differences (the Lachmannian or neo-Lachmannian approach, and the weighted phylogenetic approach) or to make no judgment at all (the unweighted phylogenetic approach). Some basis for judgment of the significance of variation is sorely needed for medieval text criticism in particular. By this, we mean that there is a need for a statistical empirical profile of the text-genealogical significance of the different sorts of variation in different sorts of medieval texts. The rules that apply to copies of Greek and Latin classics may not apply to copies of medieval Dutch story collections; the practices of copying authoritative texts such as the Bible will most likely have been different from the practices of copying the Lives of local saints and other commonly adapted texts. It is nevertheless imperative that we have a consistent, flexible, and analytically tractable model for capturing these phenomena of transmission. In this article, we present a computational model that captures most of the phenomena of text variation, and a method for analysis of one or more stemma hypotheses against the variation model. We apply this method to three ‘artificial traditions’ (i.e. texts copied under laboratory conditions by scholars to study the properties of text variation) and four genuine medieval traditions whose transmission history is known or deduced in varying degrees. Although our findings are necessarily limited by the small number of texts at our disposal, we demonstrate here some of the wide variety of calculations that can be made using our model. Certain of our results call sharply into question the utility of excluding ‘trivial’ variation such as orthographic and spelling changes from stemmatic analysis.},
	number = {4},
	urldate = {2023-08-26},
	journal = {Literary and Linguistic Computing},
	author = {Andrews, Tara L. and Macé, Caroline},
	month = dec,
	year = {2013},
	pages = {504--521},
}

Stemmatology, or the reconstruction of the transmission history of texts, is a field that stands particularly to gain from digital methods. Many scholars already take stemmatic approaches that rely heavily on computational analysis of the collated text (e.g. Robinson and O’Hara 1996; Salemans 2000; Heikkilä 2005; Windram et al. 2008 among many others). Although there is great value in computationally assisted stemmatology, providing as it does a reproducible result and allowing access to the relevant methodological process in related fields such as evolutionary biology, computational stemmatics is not without its critics. The current state-of-the-art effectively forces scholars to choose between a preconceived judgment of the significance of textual differences (the Lachmannian or neo-Lachmannian approach, and the weighted phylogenetic approach) or to make no judgment at all (the unweighted phylogenetic approach). Some basis for judgment of the significance of variation is sorely needed for medieval text criticism in particular. By this, we mean that there is a need for a statistical empirical profile of the text-genealogical significance of the different sorts of variation in different sorts of medieval texts. The rules that apply to copies of Greek and Latin classics may not apply to copies of medieval Dutch story collections; the practices of copying authoritative texts such as the Bible will most likely have been different from the practices of copying the Lives of local saints and other commonly adapted texts. It is nevertheless imperative that we have a consistent, flexible, and analytically tractable model for capturing these phenomena of transmission. In this article, we present a computational model that captures most of the phenomena of text variation, and a method for analysis of one or more stemma hypotheses against the variation model. We apply this method to three ‘artificial traditions’ (i.e. texts copied under laboratory conditions by scholars to study the properties of text variation) and four genuine medieval traditions whose transmission history is known or deduced in varying degrees. Although our findings are necessarily limited by the small number of texts at our disposal, we demonstrate here some of the wide variety of calculations that can be made using our model. Certain of our results call sharply into question the utility of excluding ‘trivial’ variation such as orthographic and spelling changes from stemmatic analysis.

@book{mcgillivray_methods_2013,
	title = {Methods in {Latin} {Computational} {Linguistics}},
	isbn = {978-90-04-26012-2},
	abstract = {In Methods in Latin Computational Linguistics, Barbara McGillivray presents some of the most significant methodological foundations of the emerging field of Latin Computational Linguistics. The reader will find an overview of the computational resources and tools available for Latin and three corpus case studies covering morpho-syntactic and lexical-semantic aspects of Latin verb valency, as well as quantitative diachronic explorations of the argument realization of Latin prefixed verbs. The computational models and the multivariate data analysis techniques employed are explained with a detailed but accessible language. Barbara McGillivray convincingly shows the challenges and opportunities of combining computational methods and historical language data, and contributes to driving the technological change that is affecting Historical Linguistics and the Humanities.},
	language = {en},
	publisher = {BRILL},
	author = {McGillivray, Barbara},
	month = nov,
	year = {2013},
	note = {Google-Books-ID: 33dfAgAAQBAJ},
	keywords = {Computers / Artificial Intelligence / Natural Language Processing, Language Arts \& Disciplines / Linguistics / Etymology, Language Arts \& Disciplines / Linguistics / General, Language Arts \& Disciplines / Linguistics / Historical \& Comparative},
}

In Methods in Latin Computational Linguistics, Barbara McGillivray presents some of the most significant methodological foundations of the emerging field of Latin Computational Linguistics. The reader will find an overview of the computational resources and tools available for Latin and three corpus case studies covering morpho-syntactic and lexical-semantic aspects of Latin verb valency, as well as quantitative diachronic explorations of the argument realization of Latin prefixed verbs. The computational models and the multivariate data analysis techniques employed are explained with a detailed but accessible language. Barbara McGillivray convincingly shows the challenges and opportunities of combining computational methods and historical language data, and contributes to driving the technological change that is affecting Historical Linguistics and the Humanities.

@article{coffee_tesserae_2013,
	title = {The {Tesserae} {Project}: intertextual analysis of {Latin} poetry},
	volume = {28},
	issn = {0268-1145, 1477-4615},
	shorttitle = {The {Tesserae} {Project}},
	url = {https://academic.oup.com/dsh/article-lookup/doi/10.1093/llc/fqs033},
	doi = {10.1093/llc/fqs033},
	abstract = {Tesserae is a web-based tool for automatically detecting allusions in Latin poetry. Although still in the start-up phase, it already is capable of identifying significant numbers of known allusions, as well as similar numbers of allusions previously unnoticed by scholars. In this article, we use the tool to examine allusions to Vergil’s Aeneid in the first book of Lucan’s Civil War. Approximately 3,000 linguistic parallels returned by the program were compared with a list of known allusions drawn from commentaries. Each was examined individually and graded for its literary significance, in order to benchmark the program’s performance. All allusions from the program and commentaries were then pooled in order to examine broad patterns in Lucan’s allusive techniques which were largely unapproachable without digital methods. Although Lucan draws relatively constantly from Vergil’s generic language in order to maintain the epic idiom, this baseline is punctuated by clusters of pointed allusions, in which Lucan frequently subverts Vergil’s original meaning. These clusters not only attend the most significant characters and events but also play a role in structuring scene transitions. Work is under way to incorporate the ability to match on word meaning, phrase context, as well as metrical and phonological features into future versions of the program.},
	language = {en},
	number = {2},
	urldate = {2023-08-26},
	journal = {Literary and Linguistic Computing},
	author = {Coffee, N. and Koenig, J.-P. and Poornima, S. and Forstall, C. W. and Ossewaarde, R. and Jacobson, S. L.},
	month = jun,
	year = {2013},
	pages = {221--228},
}

Tesserae is a web-based tool for automatically detecting allusions in Latin poetry. Although still in the start-up phase, it already is capable of identifying significant numbers of known allusions, as well as similar numbers of allusions previously unnoticed by scholars. In this article, we use the tool to examine allusions to Vergil’s Aeneid in the first book of Lucan’s Civil War. Approximately 3,000 linguistic parallels returned by the program were compared with a list of known allusions drawn from commentaries. Each was examined individually and graded for its literary significance, in order to benchmark the program’s performance. All allusions from the program and commentaries were then pooled in order to examine broad patterns in Lucan’s allusive techniques which were largely unapproachable without digital methods. Although Lucan draws relatively constantly from Vergil’s generic language in order to maintain the epic idiom, this baseline is punctuated by clusters of pointed allusions, in which Lucan frequently subverts Vergil’s original meaning. These clusters not only attend the most significant characters and events but also play a role in structuring scene transitions. Work is under way to incorporate the ability to match on word meaning, phrase context, as well as metrical and phonological features into future versions of the program.

@inproceedings{koppel_automatically_2013,
	address = {Seattle, Washington, USA},
	title = {Automatically {Identifying} {Pseudepigraphic} {Texts}},
	url = {https://aclanthology.org/D13-1151},
	urldate = {2023-07-23},
	booktitle = {Proceedings of the 2013 {Conference} on {Empirical} {Methods} in {Natural} {Language} {Processing}},
	publisher = {Association for Computational Linguistics},
	author = {Koppel, Moshe and Seidman, Shachar},
	month = oct,
	year = {2013},
	pages = {1449--1454},
}

@article{martinez_partspeech_2012,
	title = {Part‐of‐speech tagging},
	volume = {4},
	issn = {1939-5108},
	url = {https://wires.onlinelibrary.wiley.com/doi/pdfdirect/10.1002/wics.195},
	number = {1},
	journal = {Wiley Interdisciplinary Reviews: Computational Statistics},
	publisher = {Wiley Online Library},
	author = {Martinez, Angel R},
	year = {2012},
	pages = {107--113},
}

@article{crane_introduction_2012,
	title = {Introduction to the special issue on corpus and computational linguistics, philology, and the linguistic heritage of humanity},
	volume = {5},
	issn = {1556-4673},
	url = {https://doi.org/10.1145/2160165.2160166},
	doi = {10.1145/2160165.2160166},
	abstract = {The articles in this issue make two complementary assertions: first, language and linguistic sources are a key element of human cultural heritage and, second, we need to integrate the ancient goals of philology with rapidly emerging methods from fields such as Corpus and Computational Linguistics. The first 15,000,000 volumes digitized by Google contained data from more than 400 languages covering more than four thousand years of the human record. We need to develop methods to explore linguistic changes and the ideas that languages encode as these evolve and circulate over millennia and on a global scale.},
	number = {1},
	urldate = {2023-08-26},
	journal = {Journal on Computing and Cultural Heritage},
	author = {Crane, Gregory and Lüdeling, Anke},
	month = apr,
	year = {2012},
	keywords = {Computational linguistics, corpus linguistics, philology},
	pages = {1:1--1:5},
}

The articles in this issue make two complementary assertions: first, language and linguistic sources are a key element of human cultural heritage and, second, we need to integrate the ancient goals of philology with rapidly emerging methods from fields such as Corpus and Computational Linguistics. The first 15,000,000 volumes digitized by Google contained data from more than 400 languages covering more than four thousand years of the human record. We need to develop methods to explore linguistic changes and the ideas that languages encode as these evolve and circulate over millennia and on a global scale.