Research And Realization Of Mining Association Rules In Distributed System

Since the concept of association rules is proposed by Agrawal in 1993,research of mining association rules has been one of the most active aspects of dataming area. Presently, research of mining association rules in centralized system has been well done and the relative theory is becoming perfect. But mining association rules in distributed system is a topic that has just been proposed and the relative theory is not as so much. With the development of Internet and the distributed-database, a great deal of data is stored in the distributed nodes of the web and it is impossible to be stored in one single node on account of communication-efficiency and security, which makes it important to find algorithm of mining association rules in distributed system and makes the meaning of our research in the thesis.In this paper, an algorithm AprTidRec based on Apriori is proposed. Based on the algorithm of mining association rules in distributed system-CD and DD, two ways of mining association rules together with their architectures in distributed system are provided. In the end of this paper, the system of mining association rules in distributed system is implemented. The system based on C/S(client to server)mode is composed of local and global modules. After run of the local module, we get association rules based on local database and we get association rules based on global database after run of the global module.To verify the validity of the algorithm, we test the system of mining association rules by means of experimental design. The results of the experiment indicate a perfect application and a high efficiency of the algorithm.