不生成候选项目集的并行挖掘关联规则

Parallel Mining Association Rules Without Generating Candidate Item Sets

为了减少并行挖掘关联规则过程中产生的网络流量及同步次数,提高挖掘效率,在Tree-DM算法的基础上,提出了不生成候选项目集的并行关联规则挖掘算法Ptree-DM。给出了Ptree-DM算法的执行示例,并对该算法的效率进行了分析和讨论。

This paper puts forward a parallel algorithm named Ptree-DM to mine association rules from distributed transaction databases without generating candidate item sets. The remarkable characteristic of Ptree-DM is that it needs a few synchronization times and generates a little network flux. It provides an example to show how Ptree-DM works, and discusses the efficiency of Ptree-DM.