Mining Multi-Level Association Rules In Distribute Database

With the coming of rapid development of computer technology and information era, how to mine efficient knowledge from under distributed environment becomes a new topic in database research areas. Association rule mining is an important task of data mining. In this thesis, we focus on research on distributed mining multiple level associations rules, and propose several solutions and efficient distribute algorithms. The following is our main research works:1) we first analyze the present generation and specification of conceptual hierarchies and consider the characters of distributed environment .Then in this paper we generate the global conceptual hierarchy by merging the local conceptual hierarchies, use XML to specify and translate the conceptual hierarchies, and get the cross level mining information from dynamic adjusted global conceptual hierarchy.2) we introduce algorithm ML_DFPT(Multiple Level Distribute Frequent Pattern Tree) for distribute mining of frequent patterns, based on FP-growth mining, that uses only three full I/O scans of the database, eliminating the need for generating the candidate items, and reducing the communication cost. By modifying ,the algorithm ML_DFPT can also distribute mine of cross level association rules using cross level mining information.3)By analyzing the rules and its specifications ,we use a tree to represent the rules so that customer can browse the rules easily and many redundant rules are pruned. Algorithms and analyses above are implemented through an antetype mode and projects. And they are proved to be efficient and feasible.