关键词: 关联规则, 频繁项集挖掘, 可扩展性, 模式树

Abstract: Pattern tree was the most frequently used data structure in frequent item-set mining. By using Pattern tree, database could be effectively compressed into main memory, and the subsequence mining task could be completed in main memory. To make further improvement on the scalability of the algorithm, we made a further study on the Pattern tree, and proposed a new algorithm called FP-DFS based on the study. FP-DFS simplifies the mining processes through applying various operations on Pattern tree. The experiments show that FP-DFS had good efficiency in frequent item-set mining.

Key words: association rule, frequent item-set mining, scalability, Pattern tree