Research On Ontology-based Deep Web Schema Matching

With the rapid development of the Word Wide Web, the number of the online searchable databases has greatly increased. These online searchable databases which always hidden behind the query interfaces are called the Deep Web. In order to make best use of these deep web data hidden behind the query interfaces, the query interfaces of deep web need to be integrated. Toward integration over the deep web, it is need to extract the schema of the query interfaces, match the extracted query interfaces and integrate the schema of the query interfaces, then obtain a unified access query interface. The users can access the deep web from the unified query interface, which shield the details unrelated to the search. Thus the users can get the best search results.This paper focuses on the matching of query interface schemas in the deep web. Currently there already have some researches related to the matching of the query interface schemas. However, these researches emphasized on the matching process, and most of these researches only apply the information which was obtained from the query interface schemas themselves. Though many schema matching methods can be applied to match the schemas, the lack of sufficient evidence which is used to find the matching in the matching process reduces the precision of the matching. This paper introduces the concept of ontology in the process of matching the interface schema. Ontology, which demonstrates explicit specification of concepts and the relations of the concepts can be used to provide more evidence for schema matching, increases the semantic information available in the process of schema matching, achieves the purpose of finding more matching and improving the effect of the matching.It need to establish the basic models for the attributes of the query interface schema firstly, and extend the basic models of attribute by the participation of domain ontology, then obtain the extended models of attribute. The extended models of attributes are used for the matching, and achieve the purpose of increasing evidence which can be used for matching. For this reason, an algorithm which extends the basic model of attribute using domain ontology is presented in this paper. This algorithm enriches the information of synonyms and data instances for the attribute. Then more evidence is available for the process of query interface schema matching. Based on the domestic and foreign researches about the deep web query interfaces schema matching and the algorithm, a query interface schema matching framework on the extended models of attribute is proposed.This paper provides a new way of thinking for ontology-based deep web schema matching. In addition to using the extended models in the process of schema matching, the use of matching results also be studied. Finally, the experimental results show that using the method of schema matching on the extended models of attribute can get a high degree of recall and precision.