Interface Integration of Deep Web Based on Ontology

A significant amount of information in Deep Web can only be accessed through the query interface of a back-end database, instead of traversing static URL links. In order to access domain-specific databases simultaneously, it is important to construct an integration interface which allows uniform access to disparate relevant sources. Therefore, a novel method of interface integration based on ontology technique is proposed in this paper. It mainly subsumes two aspects: schema matching and schema merging. Schema matching is used to accurately identify the semantic correspondences among the attributes from different interfaces by exploiting the “bridge” effect of ontology, which can match many schemas and find all mapping relationships at one time. Schema merging is used to merge the source query interfaces to construct a unified schema based on the identified mapping relationships after schema matching, which should encompass all unique attributes features and sequences over the given set of interfaces as much as possible. Through a detailed experimental evaluation, it is indicated that the approach of interface integration based on ontology not only reduce the complexity of schema matching instead of finding pairwise-attribute correspondence in isolation, but also greatly improve the integration accuracy of interfaces. Therefore, the ontology-assisted interface integration approach is feasible and effective.