Deep Web Database Selection Of Books Area

There are a lot of online databases which are available to visit, and the different databases vary with the correlation degree of user-specific queries, therefore how to quickly and accurately find the Web databases which are closely related with the user's query becomes more and more important. This paper proposed a database selection method which is based on information retrieval.First, we improve the database selection method based on the original one, that is to say, classify the databases according to the database topic before the database selection, and therefore avoid to retrieval all the databases, which improved the access efficiency. And it is only a more detailed classification just for the books area.Second, this paper propose using of domain ontology to build text query terms, the method makes the full use of the uniqueness and attributes information of the domain ontology examples, which establishes foundation for Web database selection through text attribute.Then, the Web database selection method proposed in this paper have considered all the evaluation parameters of information retrieval performance and improved it on the basis of combined original index, at last proposed a new index of information retrieval performance and implemented the structured Web database selection. The improved method not only abandons the shortcomings of the original evaluation parameters of information retrieval performance, inherits its'advantages, but also can evaluate the performance of the multiple systems.Finally, we choose a part of the Web databases which are both efficient and accurate by determine the threshold on the basis of the sorting. For the new query word, we use FP algorithm to find the frequently queried databases, and let them be the default candidates of Web database selection.The experiments show that the Web database selection method based on information retrieval improve the efficiency of information retrieval and query accuracy to some extent.