\n
\n\n \n \n \n \n \n Lightweight LLMs for 3GPP Specifications: Fine-Tuning, Retrieval-Augmented Generation and Quantization.\n \n \n \n\n\n \n Lopes Júnior, J. D. A. P.; Pereira, J.; Do Carmo, D. S.; Lotufo, R.; and Rothenberg, C. E.\n\n\n \n\n\n\n In
2025 IEEE 11th International Conference on Network Softwarization (NetSoft), pages 37-42, 2025. \n
\n\n
\n\n
\n\n
\n\n \n\n \n \n doi\n \n \n\n \n link\n \n \n\n bibtex\n \n\n \n\n \n\n \n \n \n \n \n \n \n\n \n \n \n \n \n\n\n\n
\n
@INPROCEEDINGS{11080581,\n author={Lopes Júnior, José De Arimatéa Passos and Pereira, Jayr and Do Carmo, Diedre Santos and Lotufo, Roberto and Rothenberg, Christian Esteve},\n booktitle={2025 IEEE 11th International Conference on Network Softwarization (NetSoft)}, \n title={Lightweight LLMs for 3GPP Specifications: Fine-Tuning, Retrieval-Augmented Generation and Quantization}, \n year={2025},\n volume={},\n number={},\n pages={37-42},\n keywords={Quantization (signal);Limiting;Terminology;Computational modeling;Retrieval augmented generation;Hardware;Telecommunications;3GPP;Software development management;Communication standards;LLM;Fine-Tuning;RAG;4Bit-Quantization;3GPP},\n doi={10.1109/NetSoft64993.2025.11080581}}\n
\n
\n\n\n\n
\n
\n\n \n \n \n \n \n \n INACIA: Integrating Large Language Models in Brazilian Audit Courts: Opportunities and Challenges.\n \n \n \n \n\n\n \n Pereira, J.; Assumpcao, A.; Trecenti, J.; Airosa, L.; Lente, C.; Cléto, J.; Dobins, G.; Nogueira, R.; Mitchell, L.; and Lotufo, R.\n\n\n \n\n\n\n
Digit. Gov.: Res. Pract., 6(1). February 2025.\n
\n\n
\n\n
\n\n
\n\n \n \n
Paper\n \n \n\n \n \n doi\n \n \n\n \n link\n \n \n\n bibtex\n \n\n \n \n \n abstract \n \n\n \n\n \n \n \n \n \n \n \n\n \n \n \n \n \n \n \n \n \n \n \n\n\n\n
\n
@article{10.1145/3652951,\nauthor = {Pereira, Jayr and Assumpcao, Andre and Trecenti, Julio and Airosa, Luiz and Lente, Caio and Cl\\'{e}to, Jhonatan and Dobins, Guilherme and Nogueira, Rodrigo and Mitchell, Luis and Lotufo, Roberto},\ntitle = {INACIA: Integrating Large Language Models in Brazilian Audit Courts: Opportunities and Challenges},\nyear = {2025},\nissue_date = {March 2025},\npublisher = {Association for Computing Machinery},\naddress = {New York, NY, USA},\nvolume = {6},\nnumber = {1},\nurl = {https://doi.org/10.1145/3652951},\ndoi = {10.1145/3652951},\nabstract = {This article introduces Instru\\c{c}\\~{a}o Assistida com Intelig\\^{e}ncia Artificial (INACIA), a groundbreaking system designed to integrate Large Language Models (LLMs) into the operational framework of Brazilian Federal Court of Accounts. The system automates various stages of case analysis, including basic information extraction, admissibility examination, Periculum in mora and Fumus boni iuris analyses, and recommendations generation. Through a series of experiments, we demonstrate INACIA’s potential in extracting relevant information from case documents, evaluating its legal plausibility, and formulating propositions for judicial decision-making. Utilizing a validation dataset alongside LLMs, our evaluation methodology, to the best of our knowledge, presents a novel approach to assessing system performance, correlating highly with human judgment. These results underscore INACIA’s potential in complex legal task handling while also acknowledging the current limitations. This study discusses possible improvements and the broader implications of applying Artificial Intelligence (AI) in legal contexts, suggesting that INACIA represents a significant step toward integrating AI in legal systems globally, albeit with cautious optimism grounded in the empirical findings.},\njournal = {Digit. Gov.: Res. Pract.},\nmonth = feb,\narticleno = {6},\nnumpages = {20},\nkeywords = {Brazilian Federal Court of Accounts (TCU), Large Language Models (LLMs), Artificial Intelligence-Assisted Legal Instruction, AI in Legal Systems}\n}\n
\n
\n\n\n
\n This article introduces Instrução Assistida com Inteligência Artificial (INACIA), a groundbreaking system designed to integrate Large Language Models (LLMs) into the operational framework of Brazilian Federal Court of Accounts. The system automates various stages of case analysis, including basic information extraction, admissibility examination, Periculum in mora and Fumus boni iuris analyses, and recommendations generation. Through a series of experiments, we demonstrate INACIA’s potential in extracting relevant information from case documents, evaluating its legal plausibility, and formulating propositions for judicial decision-making. Utilizing a validation dataset alongside LLMs, our evaluation methodology, to the best of our knowledge, presents a novel approach to assessing system performance, correlating highly with human judgment. These results underscore INACIA’s potential in complex legal task handling while also acknowledging the current limitations. This study discusses possible improvements and the broader implications of applying Artificial Intelligence (AI) in legal contexts, suggesting that INACIA represents a significant step toward integrating AI in legal systems globally, albeit with cautious optimism grounded in the empirical findings.\n
\n\n\n
\n
\n\n \n \n \n \n \n \n Enhancing Brazilian Portuguese Augmentative and Alternative Communication with Card Prediction and Colourful Semantics.\n \n \n \n \n\n\n \n Pereira, J.; Rodrigues, F.; Pereira, J.; Zanchettin, C.; and Fidalgo, R.\n\n\n \n\n\n\n In
Proceedings of the 40th ACM/SIGAPP Symposium on Applied Computing, of
SAC '25, pages 19–26, New York, NY, USA, 2025. Association for Computing Machinery\n
\n\n
\n\n
\n\n
\n\n \n \n
Paper\n \n \n\n \n \n doi\n \n \n\n \n link\n \n \n\n bibtex\n \n\n \n \n \n abstract \n \n\n \n\n \n \n \n \n \n \n \n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n\n\n
\n
@inproceedings{10.1145/3672608.3707926,\nauthor = {Pereira, Jayr and Rodrigues, Francisco and Pereira, Jaylton and Zanchettin, Cleber and Fidalgo, Robson},\ntitle = {Enhancing Brazilian Portuguese Augmentative and Alternative Communication with Card Prediction and Colourful Semantics},\nyear = {2025},\nisbn = {9798400706295},\npublisher = {Association for Computing Machinery},\naddress = {New York, NY, USA},\nurl = {https://doi.org/10.1145/3672608.3707926},\ndoi = {10.1145/3672608.3707926},\nabstract = {This paper presents an approach to enhancing Augmentative and Alternative Communication (AAC) systems tailored to Brazilian Portugues by integrating Colourful Semantics (CS) with transformer-based language models. We introduce an adapted BERT model, BERTptCS, which incorporates the CS framework for improved prediction of communication cards. The primary aim is to enhance the accuracy and contextual relevance of communication card predictions, which are essential in AAC systems for individuals with complex communication needs (CCN). We compared BERTptCS with a baseline model, BERTptAAC, which lacks CS integration. Our results demonstrate that BERTptCS significantly outperforms BERTptAAC in various metrics, including top-k accuracy, Mean Reciprocal Rank (MRR), and Entropy@K. Integrating CS into the language model improves prediction accuracy and offers a more intuitive and contextual understanding of user inputs, facilitating more effective communication.},\nbooktitle = {Proceedings of the 40th ACM/SIGAPP Symposium on Applied Computing},\npages = {19–26},\nnumpages = {8},\nkeywords = {augmentative and alternative communication, colourful semantics, transformer-based language models, BERT, communication card prediction, brazilian portuguese},\nlocation = {Catania International Airport, Catania, Italy},\nseries = {SAC '25}\n}\n
\n
\n\n\n
\n This paper presents an approach to enhancing Augmentative and Alternative Communication (AAC) systems tailored to Brazilian Portugues by integrating Colourful Semantics (CS) with transformer-based language models. We introduce an adapted BERT model, BERTptCS, which incorporates the CS framework for improved prediction of communication cards. The primary aim is to enhance the accuracy and contextual relevance of communication card predictions, which are essential in AAC systems for individuals with complex communication needs (CCN). We compared BERTptCS with a baseline model, BERTptAAC, which lacks CS integration. Our results demonstrate that BERTptCS significantly outperforms BERTptAAC in various metrics, including top-k accuracy, Mean Reciprocal Rank (MRR), and Entropy@K. Integrating CS into the language model improves prediction accuracy and offers a more intuitive and contextual understanding of user inputs, facilitating more effective communication.\n
\n\n\n
\n
\n\n \n \n \n \n \n SurveySum: A Dataset for Summarizing Multiple Scientific Articles into a Survey Section.\n \n \n \n\n\n \n Fernandes, L. C.; Guedes, G. B.; Laitz, T. S.; Almeida, T. S.; Nogueira, R.; Lotufo, R.; and Pereira, J.\n\n\n \n\n\n\n In Paes, A.; and Verri, F. A. N., editor(s),
Intelligent Systems, pages 431–444, Cham, 2025. Springer Nature Switzerland\n
\n\n
\n\n
\n\n
\n\n \n\n \n\n \n link\n \n \n\n bibtex\n \n\n \n \n \n abstract \n \n\n \n\n \n \n \n \n \n \n \n\n \n \n \n\n\n\n
\n
@InProceedings{10.1007/978-3-031-79032-4_30,\nauthor="Fernandes, Leandro Car{\\'i}sio\nand Guedes, Gustavo Bartz\nand Laitz, Thiago Soares\nand Almeida, Thales Sales\nand Nogueira, Rodrigo\nand Lotufo, Roberto\nand Pereira, Jayr",\neditor="Paes, Aline\nand Verri, Filipe A. N.",\ntitle="SurveySum: A Dataset for Summarizing Multiple Scientific Articles into a Survey Section",\nbooktitle="Intelligent Systems",\nyear="2025",\npublisher="Springer Nature Switzerland",\naddress="Cham",\npages="431--444",\nabstract="Document summarization is a task to shorten texts into concise and informative summaries. This paper introduces a novel dataset designed for summarizing multiple scientific articles into a section of a survey. Our contributions are: (1) SurveySum, a new dataset addressing the gap in domain-specific summarization tools; (2) two specific pipelines to summarize scientific articles into a section of a survey; and (3) the evaluation of these pipelines using multiple metrics to compare their performance. Our results highlight the importance of high-quality retrieval stages and the impact of different configurations on the quality of generated summaries.",\nisbn="978-3-031-79032-4"\n}\n\n
\n
\n\n\n
\n Document summarization is a task to shorten texts into concise and informative summaries. This paper introduces a novel dataset designed for summarizing multiple scientific articles into a section of a survey. Our contributions are: (1) SurveySum, a new dataset addressing the gap in domain-specific summarization tools; (2) two specific pipelines to summarize scientific articles into a section of a survey; and (3) the evaluation of these pipelines using multiple metrics to compare their performance. Our results highlight the importance of high-quality retrieval stages and the impact of different configurations on the quality of generated summaries.\n
\n\n\n
\n
\n\n \n \n \n \n \n \n Aplicação de Large Language Models na Análise e Síntese de Documentos Jurídicos: Uma Revisão de Literatura.\n \n \n \n \n\n\n \n Belarmino, M.; Coelho, R.; Lotufo, R.; and Pereira, J.\n\n\n \n\n\n\n In
Anais do XIII Latin American Symposium on Digital Government, pages 193–202, Porto Alegre, RS, Brasil, 2025. SBC\n
\n\n
\n\n
\n\n
\n\n \n \n
Paper\n \n \n\n \n \n doi\n \n \n\n \n link\n \n \n\n bibtex\n \n\n \n\n \n\n \n \n \n \n \n \n \n\n \n \n \n\n\n\n
\n
@inproceedings{wcge,\n author = {Matheus Belarmino and Rackel Coelho and Roberto Lotufo and Jayr Pereira},\n title = { Aplicação de Large Language Models na Análise e Síntese de Documentos Jurídicos: Uma Revisão de Literatura},\n booktitle = {Anais do XIII Latin American Symposium on Digital Government},\n location = {Maceió/AL},\n year = {2025},\n keywords = {},\n issn = {2763-8723},\n pages = {193--202},\n publisher = {SBC},\n address = {Porto Alegre, RS, Brasil},\n doi = {10.5753/lasdigov.2025.9088},\n url = {https://sol.sbc.org.br/index.php/wcge/article/view/36331}\n}\n
\n
\n\n\n\n