The Research Of Distributed Data Mining Based On Multi-Agent

The technical development of the computer network and the distributed database leads to the storage of the huge data in a distributed way, and it promotes the development of new technology of Distributed Data Mining (DDM). DDM finds useful information with the distributed technology, and has wide application prospect.Since 1990s, M-Agent system is gradually becoming a research hotspot and has a profound impact on the development of science and technology. The M-Agent technology is suitable for the complex distributed problems. In Data Mining System based on M-Agent, Agents on different database will be working according to the user's requirements, and transferring the result to the users. So there are advantages of using the M-Agent technology in the DDM Systems.In this paper, the principle of Data Mining based on M-Agent technology is discussed. At first the development of the Agent technology and the Data Mining technology is discussed, and the benefit of their association is analyzed. Then a DDM Model based on M-Agent is established. At last, the methods in the Data Minging of Distributed Association Rules are discussed.Three parts are included in this paper: the theoretical foundation of Agent and the technology of Data Mining are discussed in the first part. Then a Data Mining Model based on M-Agent is given and the Agent cooperation is analyzed in the second part. In the last part, the methods of mining about distributed association rules are analyzed from two situations of transactional distribution and attributed distribution.