| The paper puts forward an Ontology-based data integration architecture based on Mediator /Wrapper method. By the method of the Mediator/Wrapper, the data is not really stored, but is distributed in autonomous data sources. The query is decomposed into sub-queries which can be answered by local data sources. Data sources are wrapped by wrappers, and provide the Mediator a uniform interface to execute sub-queries. The results of sub-queries are combined by the Mediator and be returned to users.Ontology as a tool is applied in the data integration system presented in this paper. Ontology can precisely express relations between concepts and sharable knowledge, which can solve semantic problems effectively. The method of establishing global Ontology and local Ontology, as well as the mapping relations, is discussed in the paper. OQL(Ontology query language) is used as the query language of the data integration system. OQL is a tree query language. By parsing the query request, a query tree can be constructed. Then by making use of the mapping rules from local Ontology to global Ontology, the query tree is decomposed into sub-trees, and the query decomposition algorithm is also presented. The sub-query is then rewritten to query language related to data sources. The paper proposed a query rewritten method to rewrite the OQL to SQL. The web Ontology Language is used to model the data. For paring and reasoning the ontology, Jena 2 development packet is made use of. Finally, the data integration system is verified by a typical test case.There are also many problems need to be researched, such as the automatic mapping between local ontology and global ontology, ontology automatic extracting and so on. |
PDF Full Text Request