Research And Application Of Comprehensive Ontology Matching Based On Dynamic Weighting Method

Ontologies considered as an effective representation of concept hierarchy andsemantic model are widely applied to many fields of computer science. IN the researchof semantic web, ontologies are considered extremely important. With the developmentof semantic web, ontology heterogeneous problems are also more and more serious. Sosolving the ontology heterogeneity problem is a key issue to the development of thesemantic web. Now, many ontology matching methods or systems appear. But thesemethods or systems still have great shortcomings such as lacking generality andscientificalness in converging the multi-strategy similarity value, or incomplete andhigh error rate in extracting the initial mappings.In order to solve the problem of ontology matching methods or systems, Acomprehensive ontology matching method based on dynamic weighting is proposedwhich combined with name, structure and attribute characteristics. First,a dynamicweighting method is proposed to solve the convergence problem. This method canautomatically compute the weighting value of each similarity strategy. Then, in order toextract a good matching from the initial similarity matrix, an improved method namedNAIVE2which is based on naive descending extraction method is proposed. This methodovercomes the error mappings of threshold problem in NAVIE and improves the mapping accuracy.In order to choose a better strategy to compute the structure’s similarity in the ontology,we use the association rule mining method which has been proved to be a good methodto measure the structure similarity.This paper use OAEI2011’s benchmark ontology test sets to do experimentalverification. Experimental results show that our framework presented in this paper isbetter. At the same time, the comprehensive ontology matching method is used in apractical project.