Conversational Knowledge Graphs

  • Larry Heck

MSR-TR-2014-70 |

The past decade has seen the emergence of web-scale structured and linked semantic knowledge graphs (KGs). These KGs provide a scalable “schema for the web,” representing a significant opportunity for the NLP and conversational-interaction (CI) research communities. This lecture describes new research that leverages KGs to bootstrap web-scale CI with no requirement for semantic schema design, no data collection, and no manual annotations. In effect, the method completes a “join” of semantic KGs to large-scale sources of NL surface forms such as Wikipedia docs and search queries. The resulting annotated data is used to bootstrap components of web-scale CI systems—speech recognizers, semantic parsers, and dialog managers. The CI system can include thousands of domains and entity types, millions of entities, and hundreds of millions of relations. The precision recall of the CI systems trained with this unsupervised method approaches those trained with supervised annotations.