Query Answering over Incomplete and Uncertain RDF

While incompleteness and uncertainty naturally arises in real-world RDF data, the RDF model itself provides little support for incomplete and uncertain data. In this paper, we introduce practical extensions of the RDF model to represent incompleteness and/or uncertainty. We adopt the semantics of certain answers as the meaningful query answers in the presence of incompleteness and uncertainty, and we investigate the computational complexity of computing the certain answers to several fragments of SPARQL under our models. We determine that the problem is hard in general, and motivated by this intractability result, we develop a heuristic algorithm for query answering in the presence of uncertain data. Our algorithm can be implemented over any RDF query evaluation engine, and it can potentially be extended to handle both incompleteness and uncertainty together.