基于马尔科夫随机场的电信欺诈用户检测方法  

Telecom Fraud Detection Method Based on Markov Random Field

作　　者：高雅诗 李静林 GAO Yashi;LI Jinglin(School of Computer Science,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京邮电大学计算机学院,北京100876

出　　处：《无线电工程》2021年第3期237-242,共6页Radio Engineering

基　　金：国家自然科学基金资助项目(61472338)。

摘　　要：电信欺诈犯罪不断出现,给社会造成了巨大的损失。针对电信欺诈用户检测,提出基于马尔科夫随机场(Markov Random Field,MRF)建立通信用户的社交网络关系模型。将用户统计行为特征与用户间的亲密度关系相结合作为网络概率特征,通过循环置信传播计算网络中各节点的后验概率作为用户的欺诈概率,完成电信欺诈用户的检测。通过在真实数据上验证,取得了较为有效的成果。With the continuous development of telecom fraud,it has caused huge losses to the society.For the detection of telecom fraud,a social network relationship model of communication users based on Markov Random Field(MRF)is established.By combining the users’statistical behavior characteristics with the users’close relationship as network probability feature,and by using the loopy belief propagation,the posterior probability of each node in the network is calculated as the user’s fraud probability.Through verification on real data,the results of this method can provide effective suggestions for telecom fraud detection and prevention.

关 键 词：电信欺诈 概率图 社交网络 置信传播 

分 类 号：TN974[电子电信—信号与信息处理]