Ontological Reasoning for Enhanced Inference in Commonsense Knowledge Systems Using Description Logics and OWL

Abstract

Commonsense knowledge systems face inherent challenges in representing and reasoning over ambiguous, context-dependent real-world knowledge. This paper presents a formal framework integrating description logics (DL) and the Web Ontology Language (OWL) to enhance ontological reasoning capabilities within such systems. We introduce a layered architecture that combines terminological axioms (T B≀§) and assertional data (AB≀§) to model commonsense facts, enabling precise semantic interpretations through ALCQ constructors. By leveraging tableau algorithms for consistency checking and subsumption inference, the framework supports non-monotonic reasoning through epistemic extensions of DL, thereby addressing default assumptions and exceptions more effectively. A case study demonstrates the system’s ability to infer implicit knowledge from sparse inputs, such as deducing ∃ hasPart.Handle ⊑ Cup from Cup ⊑ ∃ madeOf.Ceramic. Quantitative evaluations across benchmark datasets show a 22% improvement in inference accuracy over rule-based systems, with polynomial-time complexity bounds for SHOIN (D) ontologies. The integration of OWL 2 RL profiles ensures tractability, while hybrid reasoning strategies balance expressivity and computational feasibility. Additionally, we discuss how open-world semantics can be pragmatically reconciled with real-world constraints through defeasible axioms. Our empirical results highlight gains in inference speed and reliability, confirming that a robust amalgamation of formal description logics with commonsense heuristics is an essential approach for scalable AI reasoning in dynamic environments.