基于概率图的IMS固话诈骗识别模型的构建及应用分析

打开文本图片集

中图分类号：TN929.53；TP393.4 文献标志码：B 文章编码：1672-7274（2025）05-0067-03

Abstract：This paper presents an IMS landline fraud identification modelbased on probabilistic graphs.It utilizes communication social networks and combines Markov RandomFields （MRF）andLoopy Belief Propagation （LBP） algorithms.Bydynamicallyanalyzing users'callbehaviors and socialrelationships，the model achievesaccurate fraud detection.Practicalapplications demonstratethatthe modelsignificantly improves the eficiencyofIMSlandline fraud management，addressing issuessuchashigh falsepositiveratesandhighlaborcosts，therebyprovidingcrucialsupport forlaw enforcement agencies and operators in effectively combating fraud.

Keywords：probabilistic graphs;IMS landline fraud identification;Markov RandomFields; loopy belief propagation; communication social networks

近年来，电信网络诈骗呈现出多样化和复杂化的特点，传统的反诈骗手段已难以满足快速、精准的识别需求[1]。（剩余3906字）