From natural language questions to SPARQL queries: a pattern-based approach

Steinmetz, Nadine; Arning, Ann-Katrin; Sattler, Kai-Uwe GND

Linked Data knowledge bases are valuable sources of knowledge which give insights, reveal facts about various relationships and provide a large amount of metadata in well-structured form. Although the format of semantic information – namely as RDF(S) – is kept simple by representing each fact as a triple of subject, property and object, the access to the knowledge is only available using SPARQL queries on the data. Therefore, Question Answering (QA) systems provide a user-friendly way to access any type of knowledge base and especially for Linked Data sources to get insight into the semantic information. As RDF(S) knowledge bases are usually structured in the same way and provide per se semantic metadata about the contained information, we provide a novel approach that is independent from the underlying knowledge base. Thus, the main contribution of our proposed approach constitutes the simple replaceability of the underlying knowledge base. The algorithm is based on general question and query patterns and only accesses the knowledge base for the actual query generation and execution. This paper presents the proposed approach and an evaluation in comparison to state-of-the-art Linked Data approaches for challenges of QA systems.

Cite

Citation style:
From natural language questions to SPARQL queries: a pattern-based approach, 2019. . Datenbanksysteme für Business, Technologie und Web (BTW 2019). https://doi.org/10.18420/btw2019-18
Could not load citation form. Default citation form is displayed.

Rights

Use and reproduction:

Export