Research On The Key Technique Of OWL-based Ontology Integration System

As a blueprint for the next generation of web development, Semantic Web is one of hottopics in the current Web technology research. Ontology is the key technology of SemanticWeb which locates in the transition layer from document description to knowledge reasoningin the architecture of the semantic web. With the intensive development of the ontologyresearch, a large amount of domain ontologies have been developed by the researchers.However, there is no standard method for building ontologies, most researchers have differentname convention and they build the ontologies based on their academic background andresearch interests. There is difference in ontologies even though in the same domain, whichleads to the heterogeneous problems of the ontologies. Ontology integration is an effectiveway to solve the problem of the knowledge reuse and sharing between the heterogeneousontologies.As the technology of WordNet semantic dictionary is imported into the ontologyintegration system, a new method of ontology integration is proposed in the paper. First, theconcepts of the two un-integrated ontologies are positioned to the nodes of the WordNetconcept hierarchy tree. According to the characteristics of the tree structure and the feature ofthe concept description, a new comprehensive strategy of concept similarity—SSIP model isproposed based on the current similarity algorithm. In the model, the similarity base onstructures, instance and property of the concepts was separately calculated, and then acomprehensive similarity value can be obtained by composting the separate values. Thefunction of sigmoid is used in the selection of the weight, which is better than the traditionmethod of hand-selection. Based on the calculated comprehensive similarity values, combingthe relationship and relative depth of the concepts obtained by WordNet, two ontologies areintegrated into global one.In the part of the analysis and implementation of the system, two test case ontologies arebuilt by Protégé. Using the algorithm proposed in SSIP model to calculate the similarity indetail, then analyzing and comparing the result.On the basis of values obtained from theprevious step,the ontologies is integrated into one after restruction based on WordNet.Themethod in the paper is proved effectiveand correct. Finally, the semi-automatic OWL-basedintegration system is demonstrated.