摘要
针对检测社交网络中的异常用户,提出了一种基于用户基本特征的异常用户检测方法。利用GirvanNewman社区发现算法将用户分为孤立点用户和社区用户两种,结合粗糙集理论,计算用户的基本特征权重,根据特征权重计算有意义的行为特征信任值,利用特征权重和特征信任值计算用户的可信度,建立异常用户检测模型。实验结果表明,所提检测的异常用户模型适用于检测比较大的数据集,和基于内容、行为特征等传统的方法相比稳定性强、精确率和效率高。
In order to detect the abnormal users in social network, an abnormal user detection method based on the basic features of users was proposed. By using Girvan-Newman community discovery algorithm, users were divided into two kinds：isolated point users and community users, and the basic feature weights of users were calculated by combining the theory of rough set. According to the feature weights, the value of the behavior characteristic was calculated, the user＇s credibility was calculated by using the feature weight and the characteristic trust value, and then the abnormal user detection model was established. Experimental results show that the anomaly user model is suitable for the detection of large data sets, and it is more stable, more accurate and more efficient than traditional methods based on content and behavior characters.
出处
《计算机应用》
CSCD
北大核心
2017年第A02期219-224,共6页
journal of Computer Applications
基金
吉林省科技发展计划重点科技攻关项目(20150204036GX)
关键词
在线社交网络
粗糙集
异常用户检测
可信度
online social network
rough set
abnormal user detection
credibility
