LL-based query answering over RDF databases

We present a method based on top-down parsing techniques for evaluating context-free path queries on RDF Graph Databases. The syntax of the query language is based on SPARQL. The language extends SPARQL by allowing the use of non-terminal symbols of a context-free grammar to specify paths on the graph. In this manner, the language subsumes the definition of regular graph patterns present in SPARQL. Our query evaluator takes an RDF graph, a context-free grammar and a declarative query, and produces tuples of values. The query evaluator proceeds in two stages: Firstly, the RDF graph is enriched with edges representing paths which correspond to strings derived by the grammar. We show that this algorithm is correct and presents a cubic worst-case run-time complexity on the number of nodes in the graph, which is an improvement over some previous work. The second stage of the evaluator uses the produced graph to identify tuples of values defined by a declarative query. In order to validate our approach, we conducted experiments by using some popular ontologies as well as synthetic databases. We compare performance results of our method with some related work.