Topic：Query Answering over Knowledge Graph

Lecturer: Dr.Yingjie Li

Time：14:00-15:00 , December 23

Place：B101,New Main Building

Biographer：

                Dr. Yingjie Li received his PhD from Lehigh University in 2013. He is now a Knowledge Data Architect in Corporate Investment Bank of JP Morgan Chase, where he is leading the research and development activities of Knowledge Graph and Semantic Web. Before joining JP Morgan Chase, he was a Research Scientist in Samsung Information System America, where he worked on context-aware computing, user modeling, etc. His research interests include Knowledge Graph, Semantic Web, Ontology, Large-scale Data Integration and Query Answering, Semantic Searchand Semantic Web Reasoning.

Abstract：

                The Knowledge Graph is a knowledge base used by Google to enhance its search engine’s search results with semantic-search information gathered from a wide variety of sources.Currently, there are large volumes of data publicly available in the knowledge graph. These data become more tightly interrelated as the number of links in the form of mappings is also growing. Typically, these data are heterogeneous, distributed and prone to dynamic changes. In order to effectively and quickly find answers to questionsagainst such knowledge graph, this talk introduces a federated query answering system over knowledge graph. This system implements an efficient, IR-inspired inverted index named term index to integrate different data sources and determine source relevance. A tree-structure algorithm is presented to answer queries by reformulating the original conjunctive query into an AND/OR tree, generating a query execution plan on the fly and dynamically executing a bottom-up greedy source collection.Experiments conducted using synthetic data and real world data and the theoretical correctness proof of algorithms have demonstrated that this system can effectively and correctly scale to dynamic, web-scale knowledge bases.