Research On Multi-Strategy Ontology Mapping In Semantic Web

With the rapid development of Semantic Web, there are more and more ontology-based applications. An ontology is a formal, explicit specification of a shared conceptualization. The distributed nature of ontology development has led to a large number of ontologies covering overlapped domains. Therefore, it is necessary to find the mapping relationship between different ontologies to reuse and share ontology. Ontology mapping is a process of finding the semantic relationship between heterogeneous ontologies and express the relationship with fomal languages. It is important to achieve interoperation between different ontologies in semantic web.For the features of ontology such as element's concept, instance and structure, this paper proposes a multi-strategy ontology mapping approach to realize automatic ontology mapping between heterogeneous ontologies. The main research works are as follows:(1) Researching on the theroies of ontology and ontology mapping and analyzing the reasons of heterogeneous ontologies. And the similiarity of ontologies is given formally.(2) Proposing a linguistic ontology mapping strategy which calculates the similarity of ontologies using edit distance and wordnet-based approach which has been modifed.(3) Proposing an instance based mapping strategy which constructs instance using semantic enrichment and calculates the similarity of ontologies by vector space model.(4) Proposing a strutcture-based ontology mapping strategy. This method calculates the similarity of ontologies by tree kernel and similarity propagation model.(5) Combining the similarity of element pair to obtain an integrated similarity and outputing the relationship of mapping by ontology discovering algorithm.Finally, this paper designs and implements a prototype for ontolgy mapping called MSOM. And we do experiments and analysis on the prposed algorithm. Experimental results demonstrate that our solution yields a higher precision.