基于分割图集的频繁闭图挖掘算法  

Mining closed frequent graph based on partitioning graph database

作　　者：薛冰[1] 张俊峰[1] 郑超[2] 

机构地区：[1]河南城建学院计算机科学与工程系,河南平顶山467036 [2]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《计算机应用研究》2011年第1期61-64,68,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60673136);河南省重点科技攻关资助项目(092102210251)

摘　　要：为了解决大规模图集挖掘算法PartGraphMining必须重复扫描图集才能得到全部频繁子图的缺点,提出了一种改进的IPMC算法,通过hash表保存同构图的hash地址和支持度,不必重复扫描图集就可快速得到全部频繁子图,再经过少量的子图同构判断得到全部频繁闭图。在实际数据集上运行的实验结果表明它比原算法的挖掘效率有所提高。In order to solve the shortage of the PartGraphMining algorithm for mining large-scale graph databases must repeatedly scan the database that could get all frequent subgraph patterns, this paper proposed a new algorithm IPMC. It could get all frequent subgrapb patterns quickly without scanning the database repeatedly through storing graph＇ s hash address and supportting in the hash table. Furthermore, obtained all closed frequent graph patterns by the judgement of few subgraph isomorphism. The experimental result on real datasets shows that new algorithm improves the efficiency of mining.

关 键 词：大规模图集 频繁子图 子图同构 频繁闭图 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]