Research And Implementation Of Multi - Strategy Ontology Mapping Based On

With the rapid development of Semantic Web technologies in the industry, the ontology of Semantic Web technologies has been widely studied and applied in many fields. The main function of the ontology is sharing and reusing knowledge and information, enhancing the interoperability between heterogeneous systems and realizing the semantic sharing among heterogeneous Internet resources. However, there haven’t been a unified standards and specifications with the ontology creation and the development of ontology also has the characteristics of distributed, subjectivity and autonomy. these all Lead to the ontology heterogeneity and inconsistencies in the same field. As the information described by these ontology cannot be shared, the development of application of ontology was hindered. Ontology mapping is the main method to eliminate ontology heterogeneity. it can solve the problem with minimal cost and achieve the sharing and reuse of knowledge at the same time.This paper, through the profound study of the ontology mapping problem, improved the exiting ontology mapping system and model and then a more complete and effective multi-strategy ontology mapping method based on FAHP was proposed to achieve the adaptive allocation of weight between the different strategy, the main research contents of this paper are as follows:(1) The improvement of Multi-strategy combination method:in order to improve the low level of automation of current multi-strategy combination method, we use the evaluation way of fuzzy analytic hierarchy process (FAHP) analysis, which is commonly used in civil engineering qualitative analysis and quantitative analysis, to analyze the characteristic of the mapped pair and to predict the contribution of four different strategies, the language, structure, properties and the instance. According to the result that is got from different strategies, it automatically distributes the corresponding weights of each strategy and comprehensively combines those strategies. This idea does not need manual intervention. It is not only improves the level of automation of multi-strategy combination method, but also overcomes the inaccurate problem which happens when simply add those numbers together, such as the Sigmoid function, weighted average method and traditional combination method.(2) The improvement of the language-based mapping method:the similarity of language characters and the semantic similarity of the character were considered, when calculate the similarity with the knowledge base WordNet, the information of different locations of the concept nodes in the hierarchy tree and the number of sub-nodes it contains were fully used to decide the right of the node for making a more precise mapping similarity calculation.(3) The improvement of the instance-based mapping method:the typical Jaccard coefficient was used to calculate the similarity between the concepts and two other new concepts:the richness and diversity were introduced to improve the accuracy of the mapping results and solve the problem induced by the traditional methods which do not consider number of differences and likely cause the problems of distortion of the mapping result. The properties of the entity were divided into data attributes and object attributes in the process of ontology mapping to reduce workload and improve mapping efficiency.We use JAVA language to develop a prototype system with the multi-strategy ontology mapping algorithm described above. Then the International authoritative OAEI ontology data set was used to verify the effect of the prototype system. Compared to four typical mapping system FALCON, COMA, CTXMATCH and RIMOM, our experimental results proved that the method can effectively improve the recall and precision of mapping result and make a more precise mapping result.