计算机应用研究2025，Vol.42Issue(9)：2631-2636,6.DOI:10.19734/j.issn.1001-3695.2024.12.0534
基于时序图神经网络的社会团体发现
Social group discovery based on temporal graph neural network
摘要
Abstract
In social event analysis,identifying relevant social groups plays a crucial role in event governance.To address the limitations of existing group discovery approaches that neglect group influence on individual node characteristics and temporal dynamics,this study proposed a G-GCN.The model enhanced node representation by incorporating group features during indi-vidual node embedding.Recognizing the significance of temporal evolution in group discovery,it further developed a TG-GCN based on G-GCN.This extended model captured temporal variations through learning node representation changes over time,achieving cross-temporal information aggregation and converting sequential interactions into evolutionary group representations.Experiments on Yelp and Amazon datasets demonstrated 0.1 accuracy improvement in group identification,confirming TG-GCN's effectiveness.The research provides new perspectives for event governance by emphasizing the importance of tem-poral-aware node representations,offering valuable insights for dynamic social event analysis and prediction.关键词
社会团体发现/图卷积神经网络/节点表示/时序图Key words
group discovery/graph convolutional network/node representation/temporal graph分类
信息技术与安全科学引用本文复制引用
李泽,赵伟超,徐慧雯..基于时序图神经网络的社会团体发现[J].计算机应用研究,2025,42(9):2631-2636,6.基金项目
中国科学院战略性先导科技专项资助项目(XDB0500103) （XDB0500103）
国家基础学科公共科学数据中心资助项目(NBSDC-DB-02) （NBSDC-DB-02）
