Join us for expert insights
Building a Tiny Knowledge Graph with Bert and Graph Convolutions
Knowledge graphs (KGs) have become an important tool for representing knowledge and improving search results. Formally, a knowledge graph is a graph database formed from entity triples of the form (subject, relation, object) where the subject and object are nodes in the graph and the relation defines the edges. When combined with natural language understanding technology capable of generating these triples from user queries, a knowledge graph can be a fast supplement to the traditional methods employed by the search engines. In this short talk, we will show how to use Google’s Named Entity Recognition to build a tiny knowledge graph based on articles about scientific topics. To search the KG, we will use BERT to build vectors from English queries and graph convolutions to optimize the search. The result is a tiny demonstration that we hope helps to illustrate some core concepts.
Dr. Dennis Gannon is Professor Emeritus in the School of Informatics and Computing at Indiana University. From 2008 until he retired in 2015 Dennis Gannon was with Microsoft Research, most recently as the Director of Cloud Research Strategy. His previous roles at Microsoft include directing research as a member of the Cloud Computing Research Group and the Extreme Computing Group. From 1985 to 2008 Gannon was with the Department of Computer Science at Indiana University where he was Science Director for the Indiana Pervasive Technology Labs and, for six years, Chair of the Department of Computer Science. He also chaired the original planning committee for the School of Informatics. He received the School of Informatics Hermes Award in 2006. He has published over 200 scientific articles and 4 books, including Cloud Computing for Science and Engineering with Ian Foster. His current work is documented in his blog. He has a Ph.D. in Computer Science from the University of Illinois, Urbana Champaign and a Ph.D. in Mathematics from the University of California, Davis.