GRaCe: A Relaxed Approach for Graph Query Caching

Abstract

SPARQL query optimization is an important issue for RDF data stores that can benefit from the usage of caching frameworks. Most caching approaches rely on a precise match semantics, that limits the number of cache hits and, as a consequence, the potential benefit. Others propose relaxed matches for the entire query, which is precisely executed over the cached result set. In this paper, to overcome these limitations we propose GRaCe, a Graph Relaxed Caching approach for RDF data stores. GRaCe supports relaxed cache matches and a relaxed query semantics, thus increasing the number of cache hits. Experimental results show that a relaxed cache can significantly reduce query execution time in all the scenarios where a relaxed query result is tolerated.