Integration of conjunctive queries over description logics into HEX-programs

Abstract

We present cq-programs, which enhance nonmonotonic description logics (dl-) programs by conjunctive queries (CQ) and union of conjunctive queries (UCQ) over Description Logics (DL) knowledge bases, as well as disjunctive rules. dl-programs had been proposed as a powerful formalism for integrating nonmonotonic logic programming and DL reasoning on a clear semantic basis. The new cq-programs have at least two advantages. First, they offer increased expressivity by allowing general (U)CQs in the body. And second, this combination of rules and ontologies gives rise to strategies for optimizing calls to the DL-reasoner by exploiting (U)CQ facilities of the DL-reasoner. To this end, we discuss some equivalences which can be exploited for program rewriting and present respective algorithms. Experimental results for the cq-program prototype show that this can lead to significant performance improvements. Moreover, the developed optimization methods may be of general interest in the context of hybrid knowledge bases.<br />HEX-programs, which extend answer-set programming (ASP) with higher-order features and provide powerful interfacing to external computation sources, have been demonstrated to be a versatile formalism for extending the ASP paradigm. The cq-program prototype dl-plugin, which will be introduced in this work, has been developed as a plugin for dlvhex, an implementation for HEX-programs. The dl-plugin integrates ASP with description logics knowledge bases by means of external atoms.<br />For this purpose, a partial equivalence between HEX-programs and cq-programs shows that HEX-programs can serve as a host language for our new formalism, provided that only monotonic dl-atoms appear in the cq-program.