Abstract: In the process of ontology concept similarity computation, there is few about the approaches of dealing with non-isa concept relation. Considering the existence of non-isa concept relation in ontology, summarize some traditional approaches of concept similarity computation, proposed a novel approach which adapts the non-isa relation. Compute the overlay-grade of ontology directed acycline concept graph information content based on tversky model, and combined the semantic distance approach, then sumed the weighted value. Experimental results show that our approach is effectively for isa concept relation and applicable for non-isa concept relation.

Key words: NON-ISA concept relation, Ontology concept similarity, Information content, Concept graph