University of Fribourg
Date: 17 April 2023
Time: 16:00 (CET)
Title: TaxoComplete: Self-Supervised Taxonomy Completion Leveraging Position-Enhanced Semantic Matching
Abstract: Taxonomies are used to organize knowledge in many applications, including recommender systems, content browsing, or web search. With the emergence of new concepts, however, static taxonomies rapidly become obsolete as they fail to capture up-to-date knowledge. In this talk, I will first introduce a technology monitoring platform based on a dynamic taxonomy that my group and I helped design. I will explain how crowdsourcing helped us get better insight and explainability on the platform, and how we improved its results using extreme label classification. Finally, I will introduce TaxoComplete, a self-supervised taxonomy completion framework that learns the representation of nodes leveraging their position in the taxonomy, and how we substantially improved on the state of the art (doubling the quality of the results in terms of HIT@k) by encapsulating information from large neighborhoods.
Speaker Biography: Philippe Cudre-Mauroux is a Full Professor at the University of Fribourg in Switzerland, and a Council Member at the Swiss National Science Foundation (responsible for Applied Computer Science). He received his Ph.D. from the Swiss Federal Institute of Technology EPFL, where he won both the Doctorate Award and the EPFL Press Mention in 2007. Before joining the University of Fribourg, he worked on information management infrastructures at IBM Watson (NY), Microsoft Research Asia and Silicon Valley, and MIT. He recently won the Verisign Internet Infrastructures Award, a Swiss National Center in Research award, a Google Faculty Research Award, as well as a 2 million Euro grant from the European Research Council. His research interests are in next-generation, Big Data management infrastructures for non-relational data and AI.