A framework for the partial evaluation of SPARQL queries

Abstract

A framework for the partial evaluation of SPARQL queries on multiple RDF data sources, both at a local and global level, is proposed. According to the proposed approach, global evaluation of queries is accomplished by first performing local evaluation on each data source, then merging the obtained results. When merging the results, term equivalence across different sources is evaluated by looking at the context of each term. Moreover, the framework allows scoring partial answers by evaluating how much a partial answer is able to capture each concept expressed in the query. Finally, a distributed index structure is proposed that supports early pruning of useless intermediate results. © 2008 Springer-Verlag.