Back to Pre-Meeting Webinars

Speakers: Miron Livny, Shanan Peters, and Theo Rekatsinas

Title of presentation: COSMOS: An AI platform for knowledge extraction from scientific publications,

This webinar is posted on the MSM YouTube Channel

Abstract:

Scientific publications encapsulate technical knowledge and contain data and descriptions of phenomena that are necessary to parameterize and evaluate models and advance scientific research. Most workflows for locating, extracting, and aggregating data and information from scientific publications involve manual effort by experts, a slow and costly step. Our project combines xDD, one of the world’s largest and most rapidly growing collections of full-text scientific content, with high throughput computing capability and COSMOS, a micro-service oriented platform for developing and deploying advanced deep learning models for document analysis and information extraction from text, tables, figures, and equations. Our overall goal is to produce an AI technical assistant that is capable of assimilating knowledge and data relevant to scientific questions and models from across the entirety of the published literature, in real time, as it is produced. In this webinar, we provide an overview of our xDD, demonstrate the current AI capabilities of COSMOS, and outline a vision for how this system will advance the pace of scientific progress. 

Github description:  DARPA ASKE- COSMOS Platform - https://github.com/UW-COSMOS/project-docs/tree/master/presentations_reports/milestone_3

Biosketches:

Theodoros (Theo) Rekatsinas is an Assistant Professor in the Department of Computer Sciences at the University of Wisconsin- Madison. He is a member of the Database Group. He earned his Ph.D. in Computer Science from the University of Maryland and was a Moore Data Postdoctoral Fellow at Stanford University. His research interests are in data management, with a focus on data integration, data cleaning, and uncertain data. Theo’s work has been recognized with an Amazon Research Award in 2018, a Best Paper Award at SDM 2015, and the Larry S. Davis Doctoral Dissertation award in 2015.

Shanan Peters is the Dean L. Morgridge Professor in the Dept. of Geoscience at the University of Wisconsin-Madison. He earned his Ph.D. from the University of Chicago and was a University of Michigan Fellow. His research involves understanding the long-term coevolution of Earth and life, which requires compiling global databases of fossils and rocks. To this end, he has overseen technical development of manually-constructed databases, including the Paleobiology Database and Macrostrat, and has led the NSF EarthCube project called GeoDeepDive, now called xDD, a platform for text and data mining from scientific publications.

Miron Livny received a B.Sc. degree in Physics and Mathematics from the Hebrew University and M.Sc. and Ph.D. degrees in Computer Science from the Weizmann Institute of Science. He is currently the John P. Morgridge Professor of Computer Science, the director of the Center for High Throughput Computing (CHTC), is leading the HTCondor project and serves as the principal investigator and technical director of the Open Science Grid (OSG). He is a member of the scientific leadership team of the Morgridge Institute of Research where he is leading the Software Assurance Market Place (SWAMP) project and is serving as the Chief Technology Officer of the Wisconsin Institutes of Discovery.

This is part of a series of recordings for the ML-MSM Meeting, see complete list of Pre-Meeting Webinars