基于图的四叉链表存储结构的最大频繁项集挖掘算法  

The maximum frequent item set mining algorithm based on the four-fork linked storage structure

作　　者：王春华[1] 宁慧[1] 邹韵[2] 郭江鸿[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 [2]哈尔滨工程大学经济管理学院,黑龙江哈尔滨150001

出　　处：《应用科技》2013年第1期76-79,共4页Applied Science and Technology

基　　金：国家自然科学基金资助项目(60975071);黑龙江省教育厅科学技术研究资助项目(12513055)

摘　　要：虽然已有的最大频繁项集挖掘算法在结构和技术上已经做了很多改进,但还是存在挖掘速度慢、效率低的缺点,在此提出了图的四叉链表存储结构和基于该存储结构的最大频繁项集挖掘算法,该结构具有一次生成多次使用,不必耗用额外的存储空间等特点,基于该存储结构的最大频繁项集挖掘算法充分利用了该存储结构的特点以及频繁扩展集的性质,有效地减少了冗余候选集的生成,降低了串的冗余存储,将串集合间的比较转化为整型数组的比较,从而使得它比已有的最大频繁项集挖掘算法在挖掘效率上有了明显的提高,最后通过实验证明了该算法较其他已有算法效率有了较大的提高.Although a variety of improvements have been done on the existing maximum frequent item mining algorithms in terms of structures and technologies, they still suffer from low efficiency. Given these shortcomings of the existing algorithms, we propose the quad-pointer linked list structure for graph and the maximum frequent item mining algorithm based on this structure. This structure possesses once-created-multiple-used property, without the need for extra storage space. This structure property and the characteristics of the frequent extension set are utilized fully by our algorithm, which effectively reduce the redundancy for the candidate generation and storage. Besides, we convert the comparison between strings into the comparison between integer arrays, which improves the efficiency greatly for the maximum frequent item mining algorithm. Through the experiments, the efficiency of our algorithm is proved to outperform the other existing algorithms.

关 键 词：四叉链表 频繁项集 存储结构 挖掘算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]