基于改进的FP-tree最大频繁模式挖掘算法  

An algorithm for mining maximal frequent patterns based on improved FP-tree

作　　者：宁慧[1] 王素红 崔立刚[1] 郭笑语[1] 徐丽[1] 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《应用科技》2016年第2期37-43,共7页Applied Science and Technology

基　　金：国家自然科学基金项目(61201084)

摘　　要：针对FP-growth算法时空效率低的问题,提出了改进的FP-tree构造算法。该算法利用动态结点插入技术构造FP-tree,能有效减小模式树的宽度,达到压缩空间的目的;同时,该算法提高了前缀路径的共享性,提高了算法的效率。针对密集型数据的频繁模式完全集难以挖掘的问题,文中提出了IFPmax最大频繁模式挖掘算法,在改进的IFP-tree结构的基础上,利用结点的秩进行预判断,充分利用最大频繁模式的性质对已经存在的结点进行标记,有效避免了节点的冗余遍历,提高了最大频繁模式挖掘算法的效率。实验表明,在不同的基准数据集上文中提出的算法更有效,避免了节点的冗余遍历,使最大频繁模式挖掘算法效率更高。To solve the problem that FP-growth algorithm is lower in efficiency of the time and space,this paper brings an improved FP-tree construction algorithm,which diminishes breadth of FP-tree to reduce occupation space by using dynamic node insert technique to construct the FP-tree. Moreover,the algorithm improves the efficiency by increasing sharing of pefix path; to solve the problem that mining completely frequent patterns set in intensive data is very difficult,this paper proposes maximal frequent patterns mining based on IFPmax on the basis of improved FP-tree construction,which uses the rank of nodes to pre-judge,labels already existing nodes according to the nature of maximal frequent patterns,and effectively avoids redundant traversal and improves the time efficiency of the maximal frequent pattern mining algorithms. The results of experiments based on different standard data sets show the algorithm is more effective,avoiding redundant traversal and improving the time efficiency of the maximal frequent pattern mining algorithms.

关 键 词：挖掘 动态结点插入技术 IFP-tree 最大频繁模式 IFPmax 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]