一种改进的频繁子图挖掘算法  

An improved algorithm for mining frequent sub-graph

作　　者：李亮[1] 陈莉[1] 李华[2] 王珊珊[1] 张敏超[1] 

机构地区：[1]西北大学信息科学与技术学院,陕西西安710127 [2]西北大学化学分析研究所,陕西西安710069

出　　处：《计算机与应用化学》2014年第2期161-165,共5页Computers and Applied Chemistry

基　　金：国家自然科学基金(21175106)

摘　　要：在大量的图数据集合中实现目标图的精确匹配是一项相当耗时的任务,为了提高检索效率,频繁子图挖掘逐渐受到广泛的研究。使用频繁子图挖掘可以去除那些与目标图极不相似的图,这样就减小了图的数据集合,从而使目标图检索变得更为快速。FFSM算法虽是一种较为有效的频繁子图挖掘算法,但在应用中存在占用大量存储空间的缺点。本文基于FFSM算法在数据预处理的基础上,将Recomputed Embedding技术整合于FFSM算法,利用改进后的算法建立索引分类。最后将新算法应用于化学虚拟合成系统的数据处理上,实验结果证明相对于FFSM算法其获得目标化合物的速度得到了显著提高。To find the target image and achieve accurate matching in the graph data set is a time-consuming task. In order to improve the retrieval efficiency, frequent sub-graph mining has been studied extensively. The use of frequent sub-graph mining can remove those with target figure was not very similar graph, thus making the search target becomes more rapid. FFSM algorithm is an algorithm for mining frequent sub-graphs effective graph, but in the application of occupying large storage space. In this paper, based on the FFSM algorithm on the basis of data preprocessing, the Recomputed Embedding technology is integrated in the FFSM algorithm, indexing classification by using the improved algorithm. Finally, the new algorithm is used in the synthesis of chemical virtual data processing system, experimental results show that compared with FFSM algorithm for target compounds speed has been greatly improved.

关 键 词：频繁子图挖掘 Recomputed Embedding技术 FFSM算法 预处理 

分 类 号：TQ015.9[化学工程]