Automatic Ontology Mapping On Multi-Strategy

As the emergence of Semantic Web, more and more people are involved. Ontologies about related domain developed by different organizations may be largely heterogeneous.As a result, the interoperation between web applications is affected. The best way to solve the heterogeneity is by Ontology Mapping, which is to establish the corresponding semantic relations between the two elements of different ontologies, so that the ontology is fully sharing and reuse.Most methods of ontology mapping adopt the combination with multi-strategy and use all kinds of useful ontology information currently.However,there are still some deficiency about these methods:First, all kinds of ontology information is not fully used;Second,the automation degree of ontology mapping is not very high; Last,the information of application fields of Semantic Web based ontology is often vague,so current ontology mapping methods are less consider the fuzzy ontology mapping.Therefore,it is necessary to research and summarize systematically about automatic ontology mapping on multi-strategy and fuzzy ontology mapping.How to calculate ontology similarity is a critical technique in ontology mapping.In this paper, the following three aspects will be mainly finished:1.Present NV-Doc of an improved ontology similarity calculation method.Based on V-Doc (Virtual Document-based to calculate ontology similarity),which is successful applied in the Falcon-AO system.It makes fully use of information not only about entity itself in ontology and one-step neighbors,but also two-step neighbors.Experiments show that NV-Doc can outperform existing V-Doc method in terms of precision and recall on two public data sets.Although there are somewhat defect in efficiency,it also exists in the acceptable range.2. Presents a new fuzzy ontology structure similarity calculation method. Ontologies are transformed into matrix, with fuzzy value of relationship between ontology elements, calculate the structure similarity of two fuzzy ontologies in order to achieve fuzzy ontology mapping.3. Reasoning fuzzy relations between concepts of fuzzy ontology.Assuming that the membership value is transitive, each pair of reachable classes is created virtual edges in order to achieve fuzzy ontology mapping.Finally, making a conclusion that is automatic ontology mapping on multi-strategy should consider fuzzy ontology, should improve the function of sub-matcher, should improve the degree of automation, and should optimize the comprehensive strategy for each sub-matcher.