Combining description logics and Horn rules with uncertainty in ARTIGENCE

Abstract

We present ARTIGENCE, a representation language that combines description logics and Horn rules with uncertainty. ARTIGENCE capabilities go beyond the similar hybrid systems presently available, and it contains three components: a highly expressive description logic ACLNR , a set of probabilistic Horn rules and a set of ground facts. The new features described, often required in realistic application domains, can be summarized in three main points. First, we obtained a sound, complete and decidable algorithm for reasoning in ARTIGENCE knowledge base, with decidability being an important indicator that the computational complexity of the language might be essential issue for practical applications. Second, ARTIGENCE was designed not only to combine the expressive power of Horn rules and description logics, but also for its ability to deal with uncertainty. Third, we consider ACLNR as a description logic component of ARTIGENCE, which is one of the most expressive description logic with decidable inference procedures so far. We also show that the specific description logic ACLNR used in our proposed framework is not mandatory, and other decidable description logics, even their probabilistic versions can be accommodated to our framework.