有向网络兴趣社区的快速挖掘算法及其在僵尸粉检测中的应用  

A Fast Mining Algorithm for Interest Community in Directed Networks and Its Application to Detection of Zombie Fans

作　　者：王晨旭[1] 秦涛[1] 管晓宏[1,2] 周亚东[1] 

机构地区：[1]西安交通大学智能网络与网络安全教育部重点实验室,西安710049 [2]清华大学自动化系智能与网络化系统研究中心,北京100084

出　　处：《西安交通大学学报》2014年第6期7-12,共6页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(61221063;61103240;6113241);国家科技支撑计划资助项目(2011BAK08B02);中央高校基本科研业务费资助项目(2012jdhz09;xjj2011015)

摘　　要：针对传统的无向网络社区挖掘方法无法实现大规模有向网络中社区有效发现的问题,提出了一种新的有向图社区及其兴趣特征快速挖掘算法。采用贪心算法求解社区划分模块性最大化的优化问题,较好地平衡了有向图社区挖掘中准确性与有效性之间的矛盾,实现对大规模微博类有向网络社区结构的有效识别;基于发现的社区,采用tf-idf算法进一步挖掘社区用户的兴趣爱好,实现了对微博网络中兴趣小组的精确挖掘。基于新浪微博的实验结果表明:所提算法不仅可以快速有效地挖掘有向网络中的社区结构及其用户的兴趣特征,还能够准确地检测出微博网络中的僵尸粉社区,研究结果对微博系统的净化、谣言控制、网络广告的精准投放等研究具有重要的参考价值。A new fast community unfolding and interests mining algorithm is proposed to solve the problem that traditional methods cannot effectively extract communities from large-scale directed networks.A greedy algorithm is used to maximize modularity so that the tradeoff between the accuracy and efficiency in the community mining of directed networks is better balanced and its application to large scale microblog networks can be realized.The users＇ interests in the extracted community are then further mined using the tf-idf algorithm to score the most-occurred phrases in the community.Experimental results based on Sina Microblog show that the proposed algorithm can not only find out the community structures and their interests quickly,but also can uncover the zombie-fans community efficiently and accurately.These results exhibit great values for system purification,rumors control and accurate delivery of online advertising in microblog systems.

关 键 词：微博 有向图 社区挖掘 用户兴趣小组 僵尸粉 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]