No âmbito das palestras do DEI, terá lugar no dia 12 de Dezembro, na sala I-105 do departamento de Engenharia Eletrotécnica e de Computadores da FEUP, pelas 14h00, a palestra intitulada: "Fact-checking for RDF Knowledge Bases" por Diego Esteves, SDA Research, University of Bonn.
Para mais detalhes e registo, consultar: http://bit.ly/2jqLty2
Within the Lecture Series DEI, it will take place on December 12th at room I-105 of the Department of Electrical and Computer Engineering, FEUP at 14:00, the talk entitled: "Fact-checking for RDF Knowledge Bases" by Diego Esteves, SDA Research, University of Bonn.
For more details and registration, see: http://bit.ly/2jqLty2
Knowledge bases (KBs) are of utmost importance for many cutting-edge applications both in scientific research and enterprise projects. They are mainly designed to store complex structured information, which depicts facts about the world. For instance, DBPedia, YAGO, Freebase, Wikidata and Google Knowledge Graph are examples of successful KBs projects. However, only a small fraction of the world information is contained in these structured sources. Moreover, the creation process is mostly based on automated information extraction (IE) methods which are naturally error-prone.
Therefore, fact extraction and fact verification are also essential steps for Knowledge Base Population (KBP). Moreover, with this respect, ranking triples (fact ranking) based on their relevance it is another relevant task.
For instance, a person could have more than one profession or nationality, but among different possibilities, one of them often stands out. To deal with the above-introduced challenges, fact-c hecking algorithms have been proposed. In RDF KBs, they can be segmented into three different categories:
triple-plausibility, triple-validation, and triple-ranking. In this thesis, we aim at exploring state-of-the-art techniques of fact-checking applied to RDF KBs, studying its limitations and developing novel solutions to overcome current benchmark results.
Diego Esteves is a Brazilian/Portuguese computer scientist. Since 2014 he lives in Germany, where he is a Researcher and Ph.D. candidate at Rheinische Friedrich-Wilhelms-Universität Bonn. Before starting his Ph.D., he worked on IT projects for 10+ years in large companies such as Accenture, B2W Inc., Wilson Sons and BTG Pactual Investment Bank. He has vast work experience in data analysis, data integration, and machine learning. Esteves was approved for studying in several of the most prestigious federal universities and technical schools in Brazil. In 2004 he had his first hands-on experience working part-time as a trainee at the IBOPE (Brazilian Institute of Public Opinion and Statistics), immediately after obtaining his technical degree in data processing from a federal technical high school (CPII). He also holds an M.Sc. in machine learning (IME - Military Institute of Engineering, 2014), an MBA in software engineering (UFRJ - Federal University of Rio de Janeiro, 2010) and a Bac helor's degree in information systems (CEFET-RJ, 2009). His primary research topic is fact-checking, which comprehends the intersection of machine learning, natural language processing, and semantic web topics.
Divulgue os seus eventos e projetos através do e-mail: firstname.lastname@example.org. Saiba mais sobre os nossos serviços no site dos SICC