In this work, we address the application of OWL ontologies in semantic reasoning, particularly under the challenges posed by incompleteness. OWL ontologies serve as a formal knowledge representation framework that is widely utilized in domains like healthcare and bioinformatics. However, real-world ontologies are often incomplete, presenting challenges for reasoning.
We focus on a fundamental subsumption reasoning problem: given an incomplete ontology and a candidate (non-entailed) subsumption, how to determine whether the subsumption is semantically plausible and provide a logically sound explanation that includes potential missing axioms.
To tackle this reasoning problem, we propose NeurOWL, an end-to-end neuro-symbolic framework that jointly performs verification and abduction, leveraging both formally defined semantics and textual semantics through Large Language Models and ontology embeddings. We evaluate NeurOWL on real-world ontologies across various domains, demonstrating strong and robust performance.
Blogger's Review: NeurOWL showcases innovative capabilities in handling incomplete ontology reasoning by combining neural networks with symbolic reasoning. This approach not only enhances the accuracy of reasoning but also provides new solutions to the problem of ontology incompleteness, suggesting a wide range of application potential.