Research On Distributed Data Integration Model Based-on Ontology

The purpose of data integration is to provide users with uniform access patterns for the distributed data sources, shield the heterogeneity of data and implements information share among heterogeneous data resources. At present, the traditional data integration methods can not satisfy the needs of users, the semantics heterogeneous of data integration is increasingly evident. Ontology is introduced into data integration domain and can be a good description of domain knowledge concept. The semantic data integration based on ontology is a trend. The data of oil domain has been accumulated for many years, it relate to many professional categories. So there is a variety of heterogeneous data in the oil domain, it is difficult for data exchange and sharing. The challenge has taken into data integration applications of oil domain.At first, for the ontology construction, the domain ontology will be build based on data element. A domain ontology construction method based on data element register is proposed, on this basis, an automatic domain ontology construction algorithm based on the classification of data elements is achieved and the problem of domain ontology automatic construction will be solved. Secondly, by the research on the semantics recognition of the relational database schema, the rules of transform from the relational database schema to the local ontology is summarized. Based on extracting the relational schema by reading the data dictionary, the local ontology construction based on data dictionary management is achieved and the problem of local ontology automatic construction will be solved. Third, for mapping between the ontology in the data integration process, by research on extensive ontology mapping language based on XML, the mapping rules between ontology and the corresponding mapping description is summarized. As well as preliminary research on ontology bi-directional mapping, the method of the same document described in bi-directional mapping is presented. The construction of ideological mapping description is based on binary. A hybrid recursive binary tree traversal algorithm is proposed to create ontology reverse mapping expression. Finally a distributed data integration model based on the mixed ontology and Mediator / Wrapper is proposed. A prototype system for integrated model is designed and developed by web services technology and ontology technology. The result shows that the integration model and related technologies are applicable, can be used to solve heterogeneous semantic issues existed in oil domain data integration and can provide reference for other domains of data integration.