An approach for transforming keyword-based queries to SPARQL on RDF data source federations

Abstract

General public highly use keyword queries to fulfil their information needs on the Web. Semantic Web aims at transforming the Web to a format which is machine readable. RDF is the common format used in the Semantic Web to store data. Several existing approaches have proposed methods for keyword query processing on RDF data. SPARQL queries are capable of retrieving more relevant results than general keyword routing on RDF data. RDF federations are capable of connecting multiple heterogeneous data sources. Federations allow us to retrieve more complete results than querying a single source. Therefore we have introduced an approach which can transform keyword queries to SPARQL on a federation of heterogeneous sources. We utilized an ontology alignment approach for resolving vocabulary level heterogeneity. A keyword index was used to match keywords to data elements on the federation while a Path Index based approach utilized to identify suitable sub-graphs which can connect keyword elements. We were able to obtain promising results about the effectiveness of the generated queries from our proposed approach and performance of the approach.