基于二叉树编码的关联规则动态挖掘算法

A DYNAMIC MINING ALGORITHM OF ASSOCIATION RULES BASED ON BINARY TREE CODING

针对项目少、事务多的数据库关联规则挖掘问题,提出一种基于二叉树编码的关联规则动态挖掘算法。通过对应事务数据库项目建立二叉树,对应项集编码定义计数数组;对照二叉树扫描记录并计数;分析计算关联规则这几个步骤可以实现关联规则的动态挖掘。该算法充分利用了二叉树的编码特性,有效降低了I/0负载,容易实现事务的增删及数据库的划分、合并,具有较强的适用性。

To solve the problem of mining association rules for the database which has a small quantity of item sets and a large quantity of transactions, this paper proposes a dynamical algorithm based on binary tree coding. We can mine association rules through the following steps. First, setting up a binary tree suiting with item set of a database. Second, defining an array for counting which is corresponding with the item set. Then scanning and counting the transaction records. Finally, analyzing and calculating the association rules. The algorithm takes full advantages of the characteristics of binary tree coding so that it can reduce I/O workload. It is easy to add or delete records at any time. Also it is easy to divide and merge data. The algorithm shows good application prospect.