Research On Heterogeneous Data Semantic Integration Framework Based On Self-adaption Ontology

The semantic integration technology has been further developed during the process of heterogeneous data integration in current enterprises. More and more enterprises implement semantic integration by using ontology as their data conception backbone. However, with the increase of integrated data sources or changes in data environment, companies are forced to re-adjust or reconstruct the integration middleware to adapt to the current environment. Because of the frequent changes and huge amount of data have been accumulated in the data environment, manual reconstructing the integration middleware becomes a difficult task in the way of enterprise's implementation of heterogeneous data integration. Therefore, building an integration system with demand adaptivity and data source compatibility can provide good support for today's enterprises.The technology of self-adaption ontology based heterogeneous data integration provides a new way of thinking in enterprise data integration. Data source compatibility can be achieved by the formal description of data conception in ontology. And the evolution mechanism in ontology can provide basic support for building a self-adaption integration system. In this paper we proposed self-adaption ontology based heterogeneous data semantic integration framework. Utilize meta-data and instance in data source to establish the mapping relation between data source and ontology. Implement self-adaption ontology middleware by using instance-driven ontology evolution method.The main research work as following:(1) To propose a framework for self-adaption ontology-based heterogeneous data integrationThe framework can be divided into three layers. Data source connection layer is used to parse meta-data and instance to establish the mapping relation between data source and ontology. Self-adaption ontology middleware layer can detect the instances in data source and find the requirement change in the integration environment. Resource layer use REST web service to encapsulate ontologyinstances and publish the resources. (2) To propose a mechanism for semantic mapping between data source and ontology middlewareBy parsing the meta-data of database, analyzing the relationship of tables, obtaining the key constrains in columns, we designed a method for semantic mapping between relation database and ontology middleware. Based on parsing the micro format from web source, distinguish the web object from HTML, we designed a method for semantic mapping between web source and ontology middleware.(3) To propose a mechanism for instance-driven ontology evolutionBased on the semantic mapping,we proposed a mechanism for instance-driven ontology evolution and give a method that can adapt to the change requirement of data environment automatically. To guarantee the consistency of ontology, we used ontology reasoner to resolve the conflict. Finally, combine self-adaption ontology and conflict recovery to implement the function of integration middleware self-adaption.At last, we developed a prototype system according the proposed method. To verify the result, the prototype system is leveraged to integrate two different types of medical data source and verify the results. Preliminary results from prototype system show that system can integrate different type data source effectively and automatically change the integration middleware when the data environment changed.