利用加权用户关系图的谱分析探测大规模电子商务水军团体

发布时间：2018-03-27 16:47

本文选题：水军　切入点：水军团体　出处：《计算机学报》2017年04期

【摘要】：电子商务水军的识别已成为众多网络水军识别领域中较为突出的研究问题.已有电子商务水军识别研究多关注网络水军自身属性和行为特征,以发现隐藏其中的网络水军行为模式.电子商务服务急速扩张刺激以获取经济利益为目的的大规模水军傀儡账号泛滥,水军逐渐形成团体规模.大规模水军泛滥使得单独网络水军的行为更加趋向正常用户,团体内部成员间不具有明显相似性,基于特征模式识别的研究方法无法很好地发现该类电子商务水军.文中定义电子商务用户的加权用户关系图模型,分析其谱特征定位用户关系图中的异常关系结构,从而找出隐藏其后的大规模电商水军团体,并提出一种基于用户关系图模型定位大规模电商水军团体的算法.文中在两个国内外最具代表性的电子商务平台(淘宝、亚马逊)数据集上进行了大量实验,并评估了算法的不同参数定位电商水军团体的能力.实验结果表明文中提出的加权用户关系图异常结构能够很好地定位隐藏较深的大规模电子商务网络水军团体,且加权用户关系图定位电子商务网络水军团体的能力优于非加权用户关系图.
[Abstract]:The identification of E-commerce water army has become a more prominent research problem in many fields of network army identification. The research on the identification of electronic commerce water army has paid more attention to its own attributes and behavior characteristics. In order to discover the behavior pattern of the network water army hidden in it. The rapid expansion of electronic commerce service stimulates the large-scale water army puppet account to obtain the economic benefit as the goal is overflowing, The large-scale flooding of the water army makes the behavior of the individual network navy more likely to be normal, and there is no obvious similarity among the members of the group. The research method based on feature pattern recognition can not find out this kind of electronic commerce army. In this paper, the weighted user relationship graph model of electronic commerce users is defined, and its spectral features are analyzed to locate the abnormal relationship structure in the user relationship diagram. In order to find out the large scale electronic commerce and water army group hidden behind, and put forward an algorithm to locate the large scale electronic commerce water army group based on the user relation graph model. In this paper, two most representative electronic commerce platforms (Taobao, Taobao), which are the most representative electronic commerce platform at home and abroad, are proposed in this paper. Amazon has done a lot of experiments on the data set, The experimental results show that the abnormal structure of weighted user relationship graph can well locate the large scale electronic commerce military groups, which are hidden deeply in the electronic commerce network, and evaluate the ability of the algorithm to locate the electronic commerce military groups with different parameters. Moreover, the ability of weighted user relationship diagram to locate e-commerce network navy group is superior to that of non-weighted user relationship chart.
【作者单位】：北京工商大学计算机与信息工程学院;
【基金】：国家自然科学基金(61170112) 教育部人文社会科学研究青年基金(13YJC860006)资助~~
【分类号】：F724.6;TP393.09

【相似文献】