Intui2: A Prototype System for Question Answering over Linked Data

An ever increasing amount of Linked Data is made available every day. Public triple stores offer the possibility of querying hundreds of millions of triples. But this information can only be retrieved using specialized query languages like SPARQL, so for the majority of Internet users, it is still unavailable. This paper presents a prototype system aimed at streamlining the access to the information stored as RDF. The system takes as input a natural language question formulated in English and generates an equivalent SPARQL query. The mapping is based on the analysis of the syntactic patterns present in the input question. In the initial evaluation results, against the 99 questions in the QALD-3 DBpedia test set, the system provides a correct answer to 30 questions and a partial answer for another 3 questions, achieving an F-measure of 0.32.