Research Of Task-oriented Schema Matching Approach

With the development of computer science and IT, the information resources stored in all kinds of computer systems can communicate with each other. They make up of the huge Network Information Space and take on the characteristics of being domain-broad, heterogeneous, huge, dynamic, and distributed. To these characteristics existing in information sharing, this paper has proposed a kind of task-oriented information integration idea. It can integrate the information of the specific field in the information space of the wide areas, which makes the users able to get the necessary information resources in time. The goal of schema matching is to establish the mapping between two schemata, and matching is the key to realizing task-oriented integration and the central issue that this paper studies, too.The paper has proposed an approach of task-oriented schema matching which carries on research from constructing schema tree, choosing information source, matching schema and describing the results, realizing the high-efficient operation of schema matching. This paper has compared matches the approach with others, and then stated the superiority of this approach.The approach will be applied in the task-oriented information integration system. This paper has analyzed the new characteristic of such an integrated system and the demands for schema matching. The approach of schema matching changes SUDAI model into a tree-like model, which makes it easier to search for each of information classes and attributes in the course of matching; Utilizing the choice to the information source can judge relevant degree between the information source and the target schema, which can reduce matching the irrelevant information source; Having proposed a linguistics match based on matching the name and restraint of the schema, a structure match based on matching the context of the schema, and an algorithm of combining the two to provide the degree of similarity between two objects to match, and we have solved the effect that difference of schemata working on matching; Adopting XML to describe the matching results, which makes the results easy to issue and integrate.Finally, the paper has designed and realized a prototype system of the schema matching tool. The design of the tool has fully considered the need of the task-oriented integration system, so it can be transplanted to this integration system conveniently. The tool has very strong effectiveness and certain generality. This paper has designed the architecture of task-oriented integration system, and proved further with an instance the feasibility and practicability of the tool in this system.The research results of this paper will be applied to the task-oriented integration system, at the same time it has certain theoretic and practical senses to carry on research schema matching of the other domain in the complicated, dynamic environment.