Efficient Non-Monotonic Reasoning for DL-Lite with Rational Closure

Description Logics (DLs) are foundational for representing knowledge, but handling defeasible or uncertain information remains a challenge. Rational Closure (RC) is a well-established non-monotonic reasoning approach that allows for more flexible and intuitive inference in the presence of exceptions or defaults. This research investigates how RC can be effectively applied to the DL-Lite family, which are lightweight description logics known for their computational tractability. The study specifically examines two key reasoning tasks: entitlement (instance checking) and Conjunctive Query (CQ) answering within the context of DL-Lite under Rational Closure. These tasks are crucial for querying and validating knowledge bases. The core innovation presented is a novel plug-in architecture designed to integrate seamlessly with existing standard classical reasoners. This architectural approach demonstrates that performing reasoning and CQ answering under Rational Closure for DL-Lite can be achieved efficiently. The method introduces only minimal computational overhead, making it a practical solution for applications requiring non-monotonic reasoning capabilities without sacrificing performance.