Ontology-Based Semantic Matching In Active Data Warehouse

Along with the persistent and fast development of network, information technology and data warehouse, there are more and more distribute heterogeneous data sources. Although the research of heterogeneous data sources has gotten much achievement, it is still a severe problem for data warehouse and data source's semantic matching due to the existing of heterogeneity in the child-databases and data warehouse. A useful idea is the active data warehouse can resolve the semantic heterogeneity automatically and identify the data meaning accurately. So it becomes the focus of research in aboard and domestic that how to supply the semantic matching server and help the data warehouse resolve the semantic problem.Ontology_based semantic matching in data warehouse can resolve the active rules and integrate the data. Users could compute the concepts' similarity, and the system could integrate the data accurately without other mapping programs. Then the active data warehouse is a real active data warehouse.This paper summarizes the development of heterogeneous data and its integration. On the base of research background, discuss the ontology's theory, build and active data warehouse's concepts and contents relatively. Then implement the system of ontology_based semantic matching active data warehouse. By the requirement of the system and the method of building a ontology, design a ontology accordingly; On the base of elasticity matching, put forward a fitter and accurater method to judge——Probability Matching; To measure the similarities of two classes, bring forward the concept of semantic distance and implement the method of computing (including the depth distance and the length distance); then put forward two functions of similarity: one is the method of computing the length and depth distance; the other is the method of sharing the information. Then bring forward the architecture of the project, the system of semantic matching. At last we complete the design of system flow, and based on the requirement of the system, we get a fit threshold。Finally this paper discusses the further research of this system.