A Research On Context-based Ontology-matching Algorithm

With the development of Internet and people's increasing desire of information, one of the targets in web information region is to figure out how to effectively handle and utilize huge amount of information. A chief method of this research is Semantic Web, which changes mining and expressing form of traditional web technology, as well as utilizes specific knowledge base. It has surpassed the bottleneck in web mining of traditional web technology caused by the over-dependent of context matching.Ontology is one that satisfies the requirement of establishing semantic information model. It's a good way to describe the objective world by information technology. However, a single ontology can neither thoroughly describe different knowledge of objective world nor draw a consensus within the area of its description way. Therefore, much of ontology that has similar meaning exists but the language and models used have some distinctions. Heterogeneous ontology prevents information from exchange and goes against the original purpose of establishing and using ontology. One effective way to solve this problem is Ontology Mapping, and the core process of it is Ontology Matching. Most recent research results of matching algorithms are based on either context or structure, but no one is faultless.Considering characteristic of ontology matching algorithms above,"a context-based ontology matching algorithm"is proposed in this paper. The main idea of this algorithm is to match two kinds of ontology concepts via conception information and context-based environmental information. We use the algorithm to establish virtual documents, including environmental information files, logic restriction files and explanatory information files, then calculate similarity and get best matching couples of the conceptive information.A context-based ontology matching system is also designed and implemented in this paper, with Java language in Eclipse+Jena programming environment. With choosing OAEI2009 standard test case as test data source, we can get some performance results such as precision, recall and F-measure. Results show that the concept context-based ontology matching algorithm, compared to purely context-based or some structural information added ones, have strong superiority in both efficiency and accuracy of ontology matching.