Research On Some Issues Of Ontologies

With the development of the Semantic Web, ontology is fully researched which as the key factor of the Semantic Web, and many heterogeneous , changeable and low quality ontologies come out, which limit the development of the Semantic Web. With purpose of improving the development of Semantic Web and implementing knowledge reuse and share, solving the problem of dynamic update, we do a research on ontology annotation, ontology concept similarity computation, and ontology evolution.First, we introduce the fact that there are many low quality and ambiguous ontologies exist in Semantic Web, propose an ontology annotation approach based on background knowledge, and introduce an annotation algorithm based on WordNet, an annotation algorithm based on Web search engine, then design and implement an ontology annotation system based on background knowledge, and do some experiments, resulsts show that our proposed approach do better than similar approach. Second, we analyse the traditional method of ontology concept similarity, with the consideration of some ontologies with non-ISA concept relation, we propose a novel compute model based on graph model and tversky model, and combine semantic distance method, experiment shows that our computation model is applicable to the ontologies with non-ISA concept relation. At last, we discuss the issue of evolution cost constrain, propose a novel ontology evolution method based on context window, transforms an ontology evolution process into a heuristic graph searching process, combine the information within the window, search an ontology change path from the whole view in the range of context window, experiment shows our method reduce the evolution cost.