K-Extractor: Automatic Knowledge Extraction for Hybrid Question Answering

This paper describes an approach to integrate unstructured and structured data and provide a natural language query interface to the consolidated knowledge base. The approach is based on deep semantic representation expressed in RDF triples. Natural language questions are automatically converted into SPARQL queries executed against the RDF store. The approach is implemented in K-Extractor platform. Two domain collections are shown: antibiotic resistant bacteria and illicit drugs.