Computing Best Ontology Excerpts via Weighted Partial Max-SAT Solving

We consider the problem of computing best excerpts of ontologies, which are selections of a certain, small number k of axioms that best capture the knowledge regarding a given set of weighted terms. Such excerpts can be useful for ontology selection, for instance. The weights of terms are to reflect different importances of considered terms, for which we propose an incompleteness measure of excerpts based on the term weights and the notion of logical difference. We present a method to compute best k-excerpts by finding all subsumption justifications and solving a weighted partial Max-SAT problem. We demonstrate the viability of our approach with an evaluation on biomedical ontologies.