אירועים והרצאות בפקולטה למדעי המחשב ע"ש הנרי ומרילין טאוב

With millions of scientific papers coming out every year, researchers are forced to allocate their attention to increasingly narrow areas. This creates isolated “research bubbles” that limit knowledge discovery and slow down scientific progress. Toward addressing this large-scale challenge for the future of science, my work explores new paradigms for helping scientists search and discover scholarly knowledge by developing novel approaches in information retrieval and AI/ML settings and models. In this talk, I will present new retrieval capabilities allowing researchers to discover relevant ideas and directions outside their "bubbles". This is done by learning structural representations of ideas and of authors, and leveraging them for identifying relationships that spark inspirations. Motivated by this setting, I will then present a new task and models we introduced for jointly addressing the fundamental challenges of diversity, ambiguity and hierarchy of language in scientific texts. We learn to construct a graph where vertices correspond to clusters of texts referring to the same underlying concept, and directed edges reflect cluster-level hierarchy. Finally, I will also discuss new weak supervision models that can be applied to scientific discovery and more generally: Learning from comparisons between groups of instances to infer individual labels, and learning aspectual document similarity by training neural models with naturally-occuring text and graph signals. Bio: Tom Hope is a postdoctoral researcher at The Allen Institute for AI (AI2) and The University of Washington, working with Daniel Weld on accelerating scientific discovery and closely collaborating with Eric Horvitz, CSO at Microsoft. Tom completed his PhD with Dafna Shahaf at the Hebrew University of Jerusalem in January 2020. His work has received four best paper awards, appeared in top venues (PNAS, KDD, AAAI, EMNLP, NAACL, WSDM, CHI, AKBC, IEEE), and received media attention from Nature and Science on his systems for COVID-19 researchers. In parallel to his PhD Tom led an applied AI research team at Intel that published award-winning work. Tom was selected for the 2021 Global Young Scientists Summit and 2019 Heidelberg Laureate Forum, and was a member of the KDD 2020 Best Paper Selection Committee.