Agent And Ontology Based Semantic Data Integration Research

As the rapid development of computer technology and Internet, the mount of Web data increased rapidly. The information in the Internet has many characteristics such as the large mount, sierious heterogeneity, strong dynamism of information contents, and distributed autonomy of information source locations. Those characteristics bring lots of troubles to information retrieve and data sharing, therefore how to integrate the heterogeneous information sources and provide those to the user in a uniform way became a highlight research topic nowadays.The heterogeneous information contains the traditional structural relational database, object-oriented database, and the semi-structural data, even the Web information sources with different query interfaces.The distributed location means that the physical location is distributed and the integration system has no right to access the information sources and the services. Recently, the highlight research problems of information integration are shifted from solving the heterogeneities of information grammar and the information structure to eliminating the heterogeneity of the semantic layer.Semantic Web provided by Tim Berners-Lee enables the machine to understand the context of information. By Semantic Web, the information is well defined, and can be understood by computers. It is good for the interaction between people and machines. For this purpose, the ontology is needed to use, which is an expression of knowledge from some fields, and provide a specific definition method. By using the ontology information, the machines can really understand the meaning of the data, and based on the understanding, the information can be dealt with by machines directly.As the increasing mount of information on Web, the number of ontologies becomes more and more. Because of the distributed and open Web environment, the information is expressed structurally, and those ontologies are developed in the concurrency environment, that makes that although there is the useable ontology, the people still develop new ontology, and lots of ontologies of same semanteme exist. The conceptual catalogs are different, moreover the relationships among them are also different, and they may be expressed by different terms. On the other hand, the construction of ontology is tough, and it is not realistic for a user to build multiple ontologies depending on the different requirements. On many occasions, single ontology is not enough to finish the task, and the cooperation of multiple ontologies is needed. For the sake of the vacuum of uniformed specification and standard for construct ontology, the phenominens such as the redundancy of ontology, the reusing of ontology and cooperations among ontologies, and the semantic conflictions among knowledges expressed by multiple ontologies are produced. Therefore, it is important to solve the ontology integration problem. Towards the large scale isomeric data, Uniformed Data Model (UDM) is proposed, which is used for expressing the isomeric data, classification, information retrieving and management of integration. For isomeric data sources of different types, by using CORBA and UDM model, the systems of DWS, IQS, TPS, DSS, QOS and SS are developed. At the meanwhile, the demo system SORE is designed, which contains the ontology-based Personalized Query Platform (PQP), real-time monitoring user system and personalized query services.Firstly, by using the ontology of Semantic Web, the ontology-based semantic information integration is proposed, and the framework of ontology information integration is presented. Secondly, the information expression model, abstraction of data scheme, and the constructions of the local and global ontologies and the mapping are researched detailedly. Thirdly, the semantic query method and the query processing combined with the instances are discussed. Finally, the E-connections theory based ontology integration and ontology modurlirazation are discussed respectivly.Internet is distributed, dynamic and open environment, so the integration of ontology in Semantic Web is tough. Furthermore the ontology integration system is very large. The agent technique and multiple agent system is a good method to solve the problems mentioned above. The dynamic and self-manage characteristics can feed the requirements of ontology integration, and is the base of intelligent integration of ontologies. Multiple agents system is composed by the network system with multiple agents. The research problem is how the intelligent agents cooperate with each other, and what the relationships among those agents are. In this dissertation, the ontology and agent technique are discussed detailedly, and the agents-based ontology semantic integration is analyzed. At last, a correspondent multi-agent-based system (MAS) is designed and realized to deal with the ontology integration problem.