Knowledge Base Entity Lookup using Named Entity Recognition: a case study on YAGO

Identification of entities or concepts present in natural language queries is of utmost importance to semantic search applications. Identified entities may then be mapped to structured representations for machine understanding and further processing. Natural Language Processing techniques such as Named Entity Recognition plays a vital role in identification and classification of concepts present in the query. Post recognition, entities could be matched against potential subjects present in knowledge repositories. This paper presents an end to end entity lookup approach for natural language query keywords to structured knowledge base concepts. For case study implementation, Stanford NER library is used in conjunction with web semantics tools such as Apache Jena in order to map the identified entity to potential subjects available in YAGO knowledge base. Illustrations are provided for subject identification from input queries. Detailed setup and configuration of the knowledge base for SPARQL querying is also elaborated with implementation details.