Evaluating Navigational RDF Queries over the Web

Semantic Web, and its underlying data format RDF, lend themselves naturally to navigational querying due to their graph-like structure. This is particularly evident when considering RDF data on the Web, where various separately published datasets reference each other and form a giant graph known as the Web of Linked Data. And while navigational queries over singular RDF datasets are supported through SPARQL property paths, not much is known about evaluating them over Linked Data. In this paper we propose a method for evaluating property path queries over the Web based on the classical AI search algorithm A*, show its optimality in the open world setting of the Web, and test it using real world queries which access a variety of RDF datasets available online and that are not necessarily known in advance.