Study On Pattern Matching In Deep Web Data Integration

With the people's expanded requirement for the deep information from Internet, the research on the data integration in Deep Web has become an increasing widespread subject. The data integration in Deep Web has become an increasingly hot topic of the information fields, at the same time as its' precondition and basement, schema matching has the extensive application in the field of the database, such as the inquired processing in data integration in Deep Web, the definition and description of the interactive relationship among the schema integration, improving or moving schemas, the schema transform from the source of database to the data warehouses, and the exchange in the information schema is usually in e-commerce, site establishment and management, component-based development and so forth. Schema matching is a data integration of the important issues.In the needs of schema matching in the data integration in Deep Web, based on the result of research of schema matching and ontology, the role of ontology and neural network in schema matching is analyzed. This paper makes use of ontology for schema matching in data integration in Deep Web to provide an effective semantic understanding, and the neural network as the matching tool to provide an automated method.Firstly, introduce relevant knowledge of Deep Web and its related research work, then expatiate the basic information used in schema matching, and show the features of schema matching of data integration in Deep Web.Secondly, to study on Deep Web schema matching, and then propose an improved schema matching method based on ontology and based on ontology and neural network, which carried out with the matching framework, the detail of the matching process and implementing steps, focused on the work that how to change the schema of the relation database and the schema of the XML database into the ontology and the work that founding and eliminating the conflictions which encountered in the matching process.In the end, experimental verification includes three parts:the module of extraction about schema ontology; the module of neural network matching; the matching experimentation and the analyzed about the results. Experimental results demonstrate that this solution has a higher efficiency.