DL + log : tight integration of description logics and disjunctive datalog

TLDR

Integrating Description Logics with Datalog rules poses semantic and computational challenges, with full integration leading to undecidability, so prior work has imposed safeness conditions that restrict expressiveness. This paper introduces DL+log, a general framework that tightly integrates Description Logics with disjunctive Datalog. DL+log extends earlier proposals by allowing a tighter interaction between DL knowledge bases and Datalog rules, and provides algorithms that establish decidability and complexity for several Description Logics. DL+log represents the most expressive yet decidable combination of Description Logics and disjunctive Datalog rules to date.

Abstract

The integration of Description Logics and Datalog rules presents many semantic and computational problems. In particular, reasoning in a system fully integrating Description Logics knowledge bases (DL-KBs) and Datalog programs is undecidable. Many proposals have overcomed this problem through a condition that limits the interaction between the DL-KB and the Datalog rules. Such a safe integration of Description Logics and Datalog provides for systems with decidable reasoning, at the price of a strong limitation in terms of expressive power. In this paper we define DL +log, a general framework for the integration of Description Logics and disjunctive Datalog. From the knowledge representation viewpoint, DL +log extends previous proposals, since it allows for a tighter form of integration between DL-KBs and Datalog rules which overcomes the main representational limits of the approaches based on the safeness condition. From the reasoning viewpoint, we present algorithms for reasoning in DL +log, and prove decidability and complexity of reasoning in DL +log for several Description Logics. To the best of our knowledge, DL+log constitutes the most powerful decidable combination of Description Logics and disjunctive Datalog rules proposed so far.