[Colloquium] [ci-announce] Seminar: Creation, Relation and Navigation in Knowledge Networks

[Eamon Kenneth Duede]

If anyone would like to meet with Prof. Weninger in the afternoon, please let me know.

Computation Institute Presentation - Data Lunch Seminar (DLS)

Speaker: Tim Weninger, University of Notre Dame
Host:  Kyle Chard
Date:  November 20, 2015
Time: 12:00 PM - 1:00 PM
Location: The University of Chicago, Searle 240A, 5735 S. Ellis Ave.

Title: Creation, Relation and Navigation in Knowledge Networks

Abstract: In this talk we present current and ongoing work about how humans create networks of information and navigate those networks in the pursuit of knowledge. First, we will describe a mathematically principled approach that learns the inherent rules that generate any network. With these rules, we can predict the next, most probable, evolution of any network. By applying this model to the most recent knowledge graph, we are able to make broad predictions about what the knowledge graph will look like in the future.  Our second project, looks at the facts present in knowledge networks. We find that these facts are typically constructed with a subject, predicate and an object, where the subject and the object typically represent some real-world entity or idea, and the predicate represents a one-way relationship between the subject and the object. Tasks like automated fact checking are easy when a true fact-triple is present in the knowledge graph; however, if a knowledge graph is a missing some fact it could be that (1) the statement is false or (2) the statement is true, but missing from the knowledge graph. In order to rate the validity of some missing fact we have created a ``definition machine'' that automatically infers an explanation-model that interprets the asserted predicate's definition with respect to the subject and the object.

Bio:  Tim Weninger is an Assistant Professor at the University of Notre Dame where he directs the Data Science Group and is a member of the Interdisciplinary Center for Networks Science and Applications (ICENSA) and a member of the University of Chicago's MetaKnowledge Network. His research interests are in data mining, machine learning and network science. The key application of his research is to identify how humans generate, curate and search for information in the pursuit of knowledge. He uses properties of these emergent networks to reason about the nature of relatedness, membership and other abstract and physical phenomena.

Information:  Lunch will be provided



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20151119/f8f80c1d/attachment.htm