快速更新全局频繁项目集

Fast Updating of Globally Frequent Itemsets

数据挖掘中的频繁项目集更新算法研究是重要的研究课题之一.目前已有的频繁项目集更新算法主要针对单机环境,有关分布式环境下的全局频繁项目集的更新算法的研究尚不多见.为此,提出了快速更新全局频繁项目集算法(fast updating algorithm for globally frequent itemsets,简称FUAGFI).该算法主要考虑数据库记录增加时全局频繁项目集的更新情况.FUAGFI利用已建立的各局部频繁模式树(frequent pattern tree,简称FP-tree)及已挖掘的全局频繁项目集,可有效地降低网络通信量,提高全局频繁项目集的更新效率.实验结果表明,所提出的更新算法是行之有效的.

The incremental updating research of frequent itemsets is an important data mining problem in data mining fields. Many sequential algorithms are proposed for incremental updating of frequent itemsets. However, very little work is done in updating frequent itemsets in distributed environment. The algorithm FUAGFI (fast updating algorithm for globally frequent itemsets) is introduced in the case of inserting, which efficiently utilizes the created locally frequent pattern trees and the mined globally frequent itemsets. Therefore, FUAGFI uses far less communication overhead and obviously improves updating efficiency of globally frequent itemsets. Experimental results show the feasibility and effectiveness of the algorithm.