RDFMatView

Abstract

The Semantic Web as an evolution of the World Wide Web aims to create a universal medium for the exchange of semantically described data. The idea of representing this information by means of directed labelled graphs, RDF, has been widely accepted by the scientific community. However querying RDF data sets to find the desired information often is highly time consuming due to the number of comparisons that are needed. In this article we propose indexes on RDF to reduce the search space and the SPARQL query processing time. Our approach is based on materialized queries, i.e., precomputed query patterns and their occurrences in the data sets.We provide a formal definition of RDFMatView indexes for SPARQL queries, a cost model to evaluate their potential impact on query performance, and a rewriting algorithm to use indexes in SPARQL queries. We also develop and compare different approaches to integrate such indexes into an existing SPARQL query engine. Our preliminary results show that our approach can drastically decrease the query processing time in comparison to conventional query processing.