Research On Method Of Deep Web Schema Matching Based On Query Interface

With the rapid development of Internet, it brings us a flood of information. But most of these Web information is hidden in various online databases, and only accessible through the query interface, therefore it is known as Deep Web. Due to the growth speed, high quality and wide coverage of Deep Web information, it has become an important source of information. In order to allow people to use these resources easily and efficiently, it is necessary for us to integrate the Deep Web information.Deep Web schema extraction and matching on the query interface are the key to information integration. This paper takes in-depth study and research on them, and proposes corresponding algorithms and solutions, which can solve the limitations of existing methods effectively. The main work of this thesis is summarized as followings:(1) Introduce relevant knowledge of Deep Web and its situation at home and abroad, then compare and analyze the traditional schema matching and Deep Web schema matching, at last summarize the advantages and disadvantages of existing method to find new ideas and methods for schema matching.(2) The existing schema extracting methods always neglect the characteristic of query interface, regarding to this problem, we propose a new method of schema extraction for query interface based on spatial clustering. This method combines the the spatial relationship of elements, takes the minimum Euclidean distance as reference and bases on clustering algorithm to solve the problem about logic attributes extraction for Deep Web query interfaces.(3) Regarding to the low efficency of schema matching for large-scale query interface, we propose a new method of schema matching for Deep Web query interface based on association matric. This mehtod first transforms the attributes in the query interface schemas into a positive-negative association matrix, then mines group attributes by the positive association matrix and mines synonymous attributes by negative association matrix. It solves the problem about complex schema matching for Deep Web query interfaces effiectively.(4) Design and implement a field-oriented Deep Web information integration system. Finally this paper designs experiment to implement the algorithms and technology menthod, and the results shows that our methods are feasible and effective.