Research Of Query Interface Integration Mechanism In Dwiis System

With the rapid development of the global network, information on Internet is growing explosively. Deep Web sites on Internet are more and more, and the number of them is still growing rapidly. Deep Web databases provide the information to users through query interfaces, but many of them in the same domain are autonomous. It would take great effort and long time to query information in a certain domain by querying each individual relevant database. Therefore, it is necessary to integrate the many Deep Web sites of the same domain and to provide users with an integrated interface. The DWIIS System is a Deep Web information integration system, which provides users with a "one-stop" way to access the Web databases.In the implementation of the DWIIS system, this thesis raises a structure feature based integration mechanism for Deep Web query interfaces. To address the issues of differences in the interface structures, the mechanism analyzes and forms a schema tree of query interface based on structure features by the tool of query interface acquiring. Then it serializes the schema tree to get an attribute sequence of schema set, and carries out the schema matching by the order of attribute sequence to get the similarity matrix. Finally, it utilizes the matrix computation to realize the construction of integrated query interface. Consequently, this mechanism has achieved the automatic integration of Deep Web query interfaces, and can increase the maximum match.This thesis first proposes a tree based query interface model with structure features. In the process of constructing query interface schema, extracting structure features of the form is required, and then the concepts of atom attribute and complex attribute are brought in. This thesis decribes the procedure of structure feature based schema acquisition of Deep Web query interface. Secondly, it elaborates the method of schema matching, and completes the similarity computation of attributes in different schemas of the same domain. It traverses the attributes in the post order based on the structure features of schema trees to generate the attribute sequences, and then completes the similarity computation of complex attributes by the order to receive the similarity matrix. Finally, it translates the similarity matrix into the equivalence matrix, and it realizes the construction of integrated interface by equivalence matrix and matrix computation. In the process of integration, there are numerous conflicts between equivalent attributes to settle. The thesis classifies the common conflicts and raises a resolving strategy for each conflict. Eventually, integrated interface is generated through integrated schema, which completes the integration of Deep Web query interface. The experiment result shows that the technology in this thesis can be in favor of querying for Deep Web data with high speed and accuracy.