Abstract: This paper presents a multi-level association rules mining algorithm DMARM which can be used in distributed environments. It can apply different thresholds at different levels. Algorithm DMARM uses polling method to resolve the communication problem among each nodes in the distributed system. By the AND and OR operation of the set in each node algorithm DMARM works out the candidate frequent pattern and also achieves the support of the pattern at the same time. Meantime it reduces the times of scanning the database.

摘要： 提出了在分布式环境下对于每一层使用不同支持度的多层关联规则挖掘问题及其算法DMARM。该算法使用轮询方法处理分布式系统中各个节点间的通信问题，在各个节点上利用集合“或”和“与”运算，在求候选频繁模式的同时求出了模式的支持度，减少了数据库的扫描次数。

关键词: 数据挖掘, 关联规则, 分布式数据库

CLC Number: 

• TP311.133.1