Abstract: This paper focuses on the research of multi-level association rules mining, and presents a fast algorithm FAMML_FPT on frequent pattern tree. This algorithm can realize data mining not only among an identical level association rule but also among cross-level association rules. In this algorithm, conceptions of the repaired items and the cross-level repaired items are brought, which is propitious to create FP-tree from lower levels to higher levels. This algorithm can reduce the scanning times of the database and eliminate the need for generating the candidate items, which improves efficiency of data mining.

Key words: data mining, multi-level association rules, frequent pattern tree(FP-tree), repaired items

摘要： 研究了多层关联规则挖掘的理论和方法，提出了一种基于FP-tree的快速挖掘算法FAMML_FPT。该算法不仅实现了同层次关联规则的挖掘，也能实现跨层次关联规则的挖掘，其中引入了修补项、跨层修补项的概念，以便从低到高逐层建立FP-tree，有效减少了扫描数据库的次数，且不用产生大量的候选项集，提高了数据挖掘的效率。

关键词: 数据挖掘, 多层关联规则, 频繁模式树, 修补项

CLC Number: 

• TP301.6