AQUACOLD: A Novel Crowdsourced Linked Data Question Answering System. Collis, N. & Frommholz, I. In Frommholz, I., Liu, H., & Melucci, M., editors, Proceedings Second Workshop on Bridging the Gap between Information Science, Information Retrieval and Data Science (BIRDS 2021), pages 11–25, 2021. CEUR-WS.org.
AQUACOLD: A Novel Crowdsourced Linked Data Question Answering System [link]Paper  abstract   bibtex   
Question Answering (QA) systems provide answers to Natural Language (NL) questions posed by humans. The Linked Data (LD) web provides an ideal knowledge base for QA as the framework expresses structure and relationships between data which assist in question parsing. Despite this, recent attempts at NL QA over LD struggle when faced with complex questions due to the challenges in automatically parsing NL into a structured query language, forcing end users to learn languages such as SPARQLwhich can be challenging for those without a technical background. There is a need for a system which returns accurate answers to complex natural language questions over linked data, improving the accessibility of linked data search by abstracting the complexity of SPARQL whilst retaining its expressivity. This work presents AQUACOLD (Aggregated Query Understanding And Construction Over Linked Data) a novel LD QA system which harnesses the power of crowdsourcing to meet this need. AQUACOLD uses query templates built by other users to answer questions which enables the system to handle queries of significant complexity. This paper provides an overview of the system and presents the results of a technical and user evaluation against the QALD-9 benchmark. Keywords
@inproceedings{Collis2021,
	title = {{AQUACOLD}: {A} {Novel} {Crowdsourced} {Linked} {Data} {Question} {Answering} {System}},
	copyright = {All rights reserved},
	abstract = {Question Answering (QA) systems provide answers to Natural Language (NL) questions posed by humans. The Linked Data (LD) web provides an ideal knowledge base for QA as the framework expresses structure and relationships between data which assist in question parsing. Despite this, recent attempts at NL QA over LD struggle when faced with complex questions due to the challenges in automatically parsing NL into a structured query language, forcing end users to learn languages such as SPARQLwhich can be challenging for those without a technical background. There is a need for a system which returns accurate answers to complex natural language questions over linked data, improving the accessibility of linked data search by abstracting the complexity of SPARQL whilst retaining its expressivity. This work presents AQUACOLD (Aggregated Query Understanding And Construction Over Linked Data) a novel LD QA system which harnesses the power of crowdsourcing to meet this need. AQUACOLD uses query templates built by other users to answer questions which enables the system to handle queries of significant complexity. This paper provides an overview of the system and presents the results of a technical and user evaluation against the QALD-9 benchmark. Keywords},
	booktitle = {Proceedings {Second} {Workshop} on {Bridging} the {Gap} between {Information} {Science}, {Information} {Retrieval} and {Data} {Science} ({BIRDS} 2021)},
	publisher = {CEUR-WS.org},
	author = {Collis, Nicholas and Frommholz, Ingo},
	editor = {Frommholz, Ingo and Liu, Haiming and Melucci, Massimo},
	year = {2021},
	pages = {11--25},
	url_paper={https://api.zotero.org/users/9984690/publications/items/JWP4SELQ/file/view}
}

Downloads: 0