Towards answering provenance-enabled SPARQL queries over RDF data cubes

Abstract

The SPARQL 1.1 standard has made it possible to formulate analytical queries in SPARQL. While some approaches have become available for processing analytical queries on RDF data cubes, little attention has been paid to answering provenance-enabled queries over such data. Yet, considering provenance is a prerequisite to being able to validate if a query result is trustworthy. The main challenge for existing triple stores is the way provenance can be encoded in standard triple stores based on context values (named graphs). Hence, in this paper we analyze the suitability of existing triple stores for answering provenanceenabled queries on RDF data cubes, identify their shortcomings, and propose an index to handle the high number of context values that provenance encoding typically entails. Our experimental results using the Star Schema Benchmark show the feasibility and scalability of our index and query evaluation strategies.