Research And Design Of Distributed Data Mining System Based On Enterprise Service Bus

With the fast development of computer network and database technique, information takes on a new look:first accumulation is faster and update is more often; second its scale becomes huger; third the distribution range of information is more and more wider; fourth its structure is more diversified than before. So the original centralized data mining technique is unable to meet these new requirements.Distributed Data Mining (DDM) technique is proposed to solve the above problem. At present, how to design a good DDM system architecture which is platform independent and supportive for kinds of DDM algorithms is a very important aspect in DDM research field. This paper does some study based on this problem, and its main work is as follows:Firstly, it introduces the background、research status and the existing problems of DDM, and then gives the details about Web Service、Service-Oriented Architecture(SOA) and Enterprise Service Bus(ESB) technique.Secondly, based on technical analysis of existing distributed systems, the paper proposes the DDM system architecture based on ESB, which fuses the sharing and open design concept of ESB, and its mining service possesses high autonomy. The proposed system architecture is platform independent、scalable and service transparent.Then this paper takes the open source ESB software named Mule as the implementation platform, develops the DDM system based on ESB with jUDDI service registry.In the end, the paper establishes a front-end application with Flex technique, and takes some bank loans simulation data to make predicted analysis. it gives out the corresponding result and the assessments. Mining results show that the proposed data mining system architecture is feasible and it can fulfill the mining task truly and reliably.