Towards Knowledge Graph-Agnostic SPARQL Query Validation for Improving Question Answering

A Knowledge Graph Question Answering (KGQA) system needs to generate a SPARQL query over a knowledge graph (KG) that is reflecting a user's information need expressed by the given natural-language question. Yet, many of these generated queries might be completely mismatching. To deal with this problem, we developed a KG-agnostic approach that is intended to increase the KGQA quality while validating SPARQL query candidates and finally removing the incorrect ones. In this demonstration, we provide the research community a Web user interface and a RESTful API to experiment with the processing of our approach and experience the possible impact of such an approach.