Goal-Directed Module Extraction for Explaining OWL DL Entailments

Module extraction methods have proved to be effective in improving the performance of some ontology reasoning tasks, including finding justifications to explain why an entailment holds in all OWL DL ontology. However. the existing module extraction methods that compute a syntactic locality-based module for the sub-concept in a subsumption entailment, though ensuring the resulting module to preserve all justifications of the entailment, may be insufficient in improving the performance of finding all justifications. This is because a syntactic locality-based module is independent of the super-concept in a subsumption entailment and always contains all concept/role assertions. In order to extract smaller modules to further optimize finding all justifications in all OWL DL ontology, we propose a goal-directed method for extracting a module that preserves all justification of a given entailment. Experimental results oil large ontologies show that a module extracted by our method is smaller than the corresponding syntactic locality-based module, making the subsequent computation of all justifications more scalable and more efficient.