Abstract: Many researchers focus on mining association rules, in which the most important step is generating frequent items set (FIS). This paper puts forward a new method named tree-DM to mining association rules, which is based on operators of transaction tree. Tree-DM has many advantages such as: (1) it only scans transaction database one time; (2) it can find transaction set which includes frequent items set; (3) it has reasonable complexity of time; etc.

Key words: Association rules, Transaction tree, Frequent transaction tree

摘要： 关联规则挖掘问题是数据挖掘中的研究热点，该文定义了事务树等概念及相关操作，在此基础上给出了仅需扫描一次事务数据库生成关联规则的算法Tree-DM。它利用项目树记录扫描信息，通过项目树的交操作生成事务树，进而利用事务树的交操作逐步产生频繁事务树，该算法的显著特点是能在发现频繁项目集的同时发现这些频繁项目集出现在哪些事务中，并就Tree-DM的性能进行了分析。

关键词: 关联规则, 事务树, 频繁事务树