软件导刊2025，Vol.24Issue(7)：67-72,6.DOI:10.11907/rjdk.241469
基于图神经网络的在线恶意用户识别研究
Research on Online Malicious User Identification Based on Graph Neural Network
摘要
Abstract
In today's digital era,the issue of cybersecurity is becoming increasingly severe.Malicious users conduct attacks and fraudulent ac-tivities through the Internet and social networks,posing threats to personal privacy,corporate security,and social order.Traditional security methods have limited effectiveness in dealing with complex network structures and malicious behaviors.Therefore,using graph neural net-works to identify malicious users has become a focus of attention.This paper explores the application of classic GNN models(such as GCN,GAT,and GraphSAGE)in identifying malicious users,as well as methods for adding differential privacy to protect data.We conduct experi-ments on two real datasets,the experimental results demonstrate that graph neural network models outperform traditional methods.This paper provides new insights into improving the identification of malicious users,which contributes to enhancing network security and user privacy protection levels.关键词
图神经网络/在线社交网络/恶意用户检测/差分隐私Key words
graph neural networks/online social networks/malicious user detection/differential privacy分类
信息技术与安全科学引用本文复制引用
周昕元,曹冉,丁培锦,刘渊,杨凯..基于图神经网络的在线恶意用户识别研究[J].软件导刊,2025,24(7):67-72,6.基金项目
江苏省高等学校基础学科(自然科学)研究面上项目(22KJD120002) （自然科学）
江苏省大学生创新创业训练计划项目(202311117120Y) （202311117120Y）
