基于谱聚类的虚假评论群组检测  

FAKE REVIEW GROUPS DETECTION BASED ON SPECTRAL CLUSTERING

作　　者：叶子成 王帮海[1] Ye Zicheng;Wang Banghai(School of Computers,Guangdong University of Technology,Guangzhou 510006,Guangdong,China)

机构地区：[1]广东工业大学计算机学院,广东广州510006

出　　处：《计算机应用与软件》2021年第8期175-181,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(61672007)。

摘　　要：为了从数据集更有效地检测出虚假评论群组。提出一种基于谱聚类的检测算法。对数据集中的多维数据样本进行分析,确定衡量用户之间相似程度的指标;利用用户相似度指标构造一幅以用户为节点、用户之间相似度为边上权值的带权评论者图;将该图的邻接矩阵作为相似度矩阵,利用谱聚类算法对其进行群组检测,将所有用户分为15个候选群组;对检测出的候选群组进一步挖掘,分析其内部特征。将不同方法检测得到的候选群组内部特征进行比较,结果表明该算法具有更高的有效性。In order to detect the fake review groups from the data set more efficiently,this paper proposes a detection algorithm based on spectral clustering.It analyzed the multi-dimensional samples in data set to determine the indicators that measure the similarity between users.It built a weighted reviewers graph using indicators with the user as the node and the similarity as the edge’s weight,and the adjacency matrix of the graph was regarded as a similarity matrix.The spectral clustering algorithm was used to divide all users into 15 candidate groups.Finally,it further explored and analyzed the internal features of candidate groups.The internal features of candidate groups obtained by different methods were compared,and the results show that the proposed algorithm is more efficient than other algorithms.

关 键 词：谱聚类 虚假评论群组 群组检测 特征分析 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]