分布式数据库的全局频繁项目集高效更新算法

Fast Updating Algorithm for Distributed Global Frequent Itemsets

提出了快速更新全局频繁项目集的算法IUAGFI(IncrementalUpdatingAlgorithmforGlobalFrequentItemsets)。该算法主要考虑数据库记录发生变化时全局频繁项目集的更新情况,在最坏的情况下仅需扫描各局部数据库一遍,并利用已建立的各局部改进的频繁模式树和已挖掘的结果,可避免传送某些原全局频繁项目对应的被约束子树,从而降低网络通讯代价。实验结果表明,该算法是有效可行的。

In this paper,a new algorithm IUAGFI（Incremental Updating Algorithm for Global Frequent Itemsets） is introduced,it considers the change of global frequent itemsets when dynamically changing database records.In the worst case,IUAGFI only scans every local transaction database once,and can avoid transmitting some constrained tree of original global frequent item by utilizing the created local improved frequent pattern tree and mined results.Therefore, IUAGFI uses far less communication overhead and obviously improves updating efficiency of global frequent itemsets. Experimental results show that IUAGFI is efficient and effective.