The Research For Data Mining Based On Multi-Agent Interation

Network technology, database and data warehouse technology enables rapid development of large accumulation of data, data from a large number of hidden knowledge found in information, data mining technology to appear. With the rapid growth of data and people to the urgent needs of tacit knowledge, data mining technology to rapid development. Distributed Data Mining using distributed computing technology, discovers that users need to focus on knowledge from the distribution database,data warehouse and other data, providing information for the user's decision-making support, has great application prospect. This value is based on the needs of the mass analysis and data mining technology development, the careful study of the Agent and Multi Agent Technology, this article in between Agent and data mining to find a valid entry point.Since 1990s, Agent and Multi Agent Technology has become the focus of research scholars, and its new ideas and solution scheme to many scientific and technological development had a profound influence. As Agent with autonomy, initiative, interaction, and adaptability for the environment etc. technical characteristics, multiple independent Agent together to complete the idea of distributed data mining tasks will naturally emerge in the minds of scholars. Thus, integrated management of multiple Agent Multi-Agent technology is very suitable for solving complex distributed problems, and data mining technology from centralized toward distributed data mining is an inevitable trend. This article is based on this idea, designs a data mining system based on multi-Agent collaborative interaction, data mining Agent of system distribution in the different regions in the network data sets, and each site's data mining Agent makes corresponding Data Mining according to the UAgent assigned tasks, UAgent makes data mining under the management and administrative control of the Global Data Mining Agent (ManagerAgent), the results firstly are sent to the Uagent which directly manages the Agent of data mining, then send it back to ManagerAgent, controls and manages the result set by the ManagerAgent or deliveries the contented and marshalled result set to the user. Uagent and ManagerAgent of the system in result set handling is a complex process, specifically how to control the system are detailed in the achieve resolution chapter. This system will avoid mining the data of different sites to be mobile, and can efficiently implement distributed parallel computing.Paper first summarizes the developmen status of the current data mining technology, Agent technology and multi-Agent technology; then furtherly analyzes the data mining technology trends; after combined the superiority of the multi-Agent technology, through multiple Agent technology and distributed data mining techniques complement each other, designs a data mining system structure based on the multi-Agent collaborative interaction; after,complets the pseudo-code implention of all function modules based on the structure;finally, uses association rule mining method to simulate the experimental test of the system and analyze the results.