极大频繁模式挖掘算法  

Algorithm for mining maximal frequent patterns

作　　者：唐德权[1] 刘绪崇[1] 姚婷婷[1] TANG De-quan;LIU Xu-chong;YAO Ting-ting(Department of Information Technology,Hunan Police Academy,Changsha 410138,China)

机构地区：[1]湖南警察学院信息技术系,湖南长沙410138

出　　处：《计算机工程与设计》2023年第6期1758-1764,共7页Computer Engineering and Design

基　　金：湖南省教育科学“十四五”规划课题基金项目(XJK23BGD034);湖南警察学院高层次人才科研启动专项基金项目(2022KYQD03);国家自然科学基金项目(61471169);湖南省科技重大专项基金项目(2017SK1040);湖南省教育厅重点基金项目(20A172)。

摘　　要：为从半结构化和结构化数据集中避免挖掘大量冗余候选模式,提高在大型图数据集中挖掘完整频繁子图的效率,提出基于极大频繁子树挖掘的算法。挖掘图数据集中所有极大频繁子树,在此基础上添加频繁边,进一步扩展操作得到所有极大频繁子图。提出定理证明极大频繁子图挖掘算法的正确性,并证明其时间复杂度优于同类挖掘算法。通过化学分子数据集、模拟数据集和大型数据集的实验验证了该算法的正确性和有效性。To avoid mining a large number of redundant candidate patterns from semi-structured and structured data sets and improve the efficiency of mining complete frequent subgraphs in large graph data sets,a method based on maximal frequent subtree mining algorithm was proposed.All maximal frequent subtrees were mined from the graph data set,and on this basis,frequent edges were added to extended operation to obtain all maximal frequent subgraphs.A theorem was proposed to prove the correctness of the maximal frequent subgraph mining algorithm,and it also proved that its time complexity is superior to the similar mining algorithms.Results of experiments on chemical molecular data sets,simulation data sets and large data sets verify the correctness and effectiveness of the algorithm.

关 键 词：图数据集 冗余子图 候选模式 频繁子图 极大频繁子树 扩展操作 极大频繁子图 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]