RDF Keyword Search Using a Type-based Summary

Keyword search enjoys great popularity due to succinctness and easy operability for exploring RDF data. SPARQL has been recommended as the standard query language that can retrieve any answers users need from available RDF data. Thus, keyword search based on keywords-to-SPARQL attracts more and more attention. However, existing solutions have main limitations that the summary index used for translation is incomplete and thus results returned are wrong or short of some answers. To address these issues, we propose an original RDF keyword search paradigm based on the translation of keyword queries to SPARQL queries. We present a new type-based summary which summarizes all the inter-entity relationships from RDF data in order to prune search space. We exploit an efficient search algorithm to quickly find the top-k subgraphs connecting all entering keyword elements. Then, a transforming algorithm is leveraged to translate the top-k subgraph into top-k SPARQL queries that are eventually executed by a SPARQL query engine. The experiments show that our approach takes shorter query response time and more accurate results are achieved than the existing techniques.