Abstract: Incomplete pruning leads to inefficiencies for the search space with the same transaction set in parent-child node according to the traditional pruning strategy. This paper presents a parent-child equivalency pruning strategy which can prune the search space with the same transaction set in parent-child node. It effectively minimizes the search space and reduces the number of frequent items. The new maximal frequent itemset mining algorithm is completed based on new pruning strategy. Experimental results show that the new pruning strategy can shorten the time of mining maximal frequent itemset with the same support.

Key words: data mining, maximum frequent itemset, pruning strategy, minimum support, depth-first, association rule

摘要： 在传统剪枝策略中，具有相同事务集的父子结点搜索空间没有充分剪枝，效率较低。为此，提出父子等价的剪枝策略。采用深度优先搜索集合枚举树，对于父子结点中具有相同事务集的搜索空间进行剪枝，有效地缩小搜索空间，减少频繁项计算的次数，给出基于该剪枝策略的最大频繁项集挖掘算法。实验结果表明，该算法可缩短同一支持度下的最大频繁项集挖掘时间。

关键词: 数据挖掘, 最大频繁项集, 剪枝策略, 最小支持度, 深度优先, 关联规则

CLC Number: 

• TP311