An optimised Semantic Web query language implementation in Prolog

Abstract. The Semantic Web is a rapidly growing research area aiming at the exchange of semantic information over the World Wide Web. The Semantic Web is built on top of RDF, an XML-based exchange language representing a triple-based data model. Higher languages such as the description logic based OWL language family are defined on top of RDF. Making inferences over triple collections is a promising application area for Prolog. In this article we study query translation and optimization in the context of the SeRQL RDF query language. Queries are translated to Prolog goals, which are optimised by reordering literals. We study the domain specific issues of this general problem. Conjunctions are often large, but the danger of poor performance of the optimiser can be avoided by exploiting the nature of the triple store. We discuss the optimisation algorithms as well as the information required from the low level storage engine.