Efficient Query Answering over Datalog with Existential Quantifiers

This paper faces the problem of answering conjunctive queries over Datalog programs allowing existential quantifiers in rule heads. Such an extension of Datalog is highly expressive, enables easy yet powerful ontology-modelling, but leads to undecidable query answering in general. To overcome undecidability, we first define Shy, a subclass of Datalog with existential quantifiers preserving not only decidability but also the same complexity of query answering over Datalog. Next, we design and implement a bottom-up evaluation strategy for Shy programs. Our computation strategy includes a number of optimizations resulting in DLV∃, a powerful reasoner over Shy programs. Finally, we carry out an experimental analysis comparing DLV∃ with some state-of-the-art systems for ontology-based query answering. The results confirm the effectiveness of DLV∃, which outperforms all other systems in the benchmark.