Research On Holistic Schema Matching Technology On Query Interface

Schema matching on web query interfaces is a difficult research and breakthrough point of the deep web databases integration, and the technology of holistic schema matching has become a hot subject of schema matching on web query interfaces. The technology of holistic schema matching can match all the schemas in the same domain at the same time. Therefore, it can greatly improve the efficiency of matching. In this paper, based on the research of the existing situation about holistic schema matching at home and abroad, the depth study is carried on the technology of query interface holistic schema matching.Firstly, for the character matching problem in pre-processing stage for the attributes, an improved APrimeAtt algorithm is proposed. This algorithm introduces the attribute weight of thematic data set to the process of primed attribute, so that algorithm is not dependent on the threshold and can ensure the attribute set stability. In addition, the algorithm is proved in theory and instance.Secondly, for the group attributes mining and synonym matching discovery problem in schema matching stage, an improved GroupAttMining algorithm and an improved Discoverysynonym algorithm are proposed. These algorithms are based on the formulas of attributes of positive association and negative association to achieve the complex schema matching and simple schema matching. In addition, these algorithms are proved in theory and instance.Thirdly, for concept partition building problem in holistic schema matching, the concept of the optimized version of concept partition and the optimal mode of model are proposed, and the problem concept partition and holistic schema matching both are proved to be NP-complete. Based on above, the Concept-partition algorithm and HSM algorithm are proposed to form concept partition and to realize holistic schema matching. In addition, these algorithms are proved in theory and instance.Finally, this algorithms which is proposed in this paper are validated througth experiment. The figures of result are given, and the results are analyzed and compared. The correctness of the algorithm is proved.