PAC learning of concept inclusions for ontology-mediated query answering
Sergei A. Obiedkov, Barış Sertkaya · International Journal of Approximate Reasoning · 2025
We present a probably approximately correct algorithm for learning the terminological part of a description-logic knowledge base via subsumption queries. The axioms we learn are concept inclusions between conjunctions of concepts from a specified set of concept descriptions. By varying the distribution of queries posed to the oracle, we adapt the algorithm to improve the recall when using the resulting TBox for ontology-mediated query answering. Experimental evaluation on OWL 2 EL ontologies suggests that our approach helps significantly improve recall while maintaining a high precision of query answering. • A PAC algorithm for learning DL ontologies via subsumption queries. • A method to fine-tune query distribution during learning to boost recall in ontology-mediated query answering. • Experimental evaluation.