Recently, Google selected Kang Liu as a winner of the Google Focused Research Award 2015, and generously supported his research on knowledge representing and reasoning. The accepted project is Representing and Reasoning Knowledge by Jointly Learning on Knowledge Graphs, Textual Triples and Unstructured Texts.
As part of Google’s ongoing commitment to support ambitious research in computer science and engineering, Google Focused Research Awards Program is for research in areas of study which is of key interest to Google, as well as the research community. Kang Liu’s accepted proposal is in the focused plan of “Google Language Understanding and Knowledge Discovery Focused Research Awards Program” which aims to identify and support world-class, full-time faculty pursuing research in areas of language understanding and knowledge discovery. Kang Liu’s proposal mainly focuses on the following two aspects:
(1) Representing knowledge by jointly learning on knowledge graphs, textual triples and unstructured texts. In this way, textual triples and texts could be regarded as supplements to KG embedding models and provide sufficient instances to learn more precise embeddings of knowledge. More importantly, they should have sufficient generalization capability to infer new entities and relations out of KG.
(2) Learning inference rules through deep learning to infer new knowledge in a large-scale scene.