With the development of more expressive description logics (DLs) for the Web Ontology Language OWL the question arises how we can properly deal with the high computational complexity for efficient reasoning. In application cases that require scalable reasoning with expressive ontologies, non-standard reasoning solutions such as approximate reasoning are necessary to tackle the intractability of reasoning in expressive DLs. In this paper, we are concerned with the approximation of the reasoning task of instance retrieval on DL knowledge bases, trading correctness of retrieval results for gain of speed. We introduce our notion of an approximate concept extension and we provide implementations to compute an approximate answer for a concept query by a suitable mapping to efficient database operations. Furthermore, we report on experiments of our approach on instance retrieval with the Wine ontology and discuss first results in terms of error rate and speed-up.
& Hitzler, P.
(2010). Approximate Instance Retrieval on Ontologies. Lecture Notes in Computer Science, 6261, 503-511.