SPARQLog: SPARQL with Rules and Quantification

Abstract

SPARQL has become the gold-standard for RDF query languages. Nevertheless, we believe there is further room for improving RDF query languages. In this chapter, we investigate the addition of rules and quantifier alternation to SPARQL. That extension, called SPARQLog, extends previous RDF query languages by arbitrary quantifier alternation: blank nodes may occur in the scope of all, some, or none of the universal variables of a rule. In addition, SPARQLog is aware of important RDF features such as the distinction between blank nodes, literals and IRIs or the RDFS vocabulary. The semantics of SPARQLog is closed (every answer is an RDF graph), but lifts RDF’s restrictions on literal and blank node occurrences for intermediary data. We show how to define a sound and complete operational semantics that can be implemented using existing logic programming techniques. While SPARQLog is Turing complete, we identify a decidable (in fact, polynomial time) fragment SwARQLog ensuring polynomial data-complexity inspired from the notion of super-weak acyclicity in data exchange. Furthermore, we prove that SPARQLog with no universal quantifiers in the scope of existential ones (∀ ∃ fragment) is equivalent to full SPARQLog in presence of graph projection. Thus, the convenience of arbitrary quantifier alternation comes, in fact, for free. These results, though here presented in the context of RDF querying, apply similarly also in the more general setting of data exchange.