《计算机应用研究》|Application Research of Computers

基于图聚类与蚁群算法的社交网络聚类算法

Clustering algorithm of social network based on graph clustering and ant colony optimization algorithm

免费全文下载 (已被下载 次)  
获取PDF全文
作者 叶小莺,万梅,唐蓉,谢云,陈桂宏,李强
机构 1.广东东软学院 计算机科学与技术系,广东 佛山 528225;2.广州工商学院 计算机科学与工程系,广州 510850;3.重庆市九龙坡区精神卫生中心,重庆 400052;4.中山大学 电子与信息工程学院,广州 510006
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)06-013-1670-05
DOI 10.19734/j.issn.1001-3695.2018.12.0881
摘要 针对社交网络中社交关系的有向性与多样性,提出了一种基于图聚类与蚁群算法的社交网络聚类算法。首先,在网络覆盖率的约束下为社交网络建立有向、非全连接的二维图模型;然后,采用K-medoids算法搜索用户分组的中心用户,采用人工蚁群算法在2D图中搜索各个用户与中心用户的相似性,将满足相似性阈值的用户分为同一个用户组。设计了低活跃用户的预测机制解决网络的稀疏性问题与冷启动问题。此外,通过网络覆盖率的约束条件权衡聚类准确率与覆盖率两个指标。仿真实验结果表明,该算法实现了较好的社交网络聚类性能,并且有效地缓解了稀疏性问题与冷启动问题。
关键词 社交网络; 数据挖掘; 聚类处理; 人工蚁群优化; 图聚类; 信任信息
基金项目 广东省科技计划协同创新与平台环境建设基金资助项目(2017A040406001)
广东省教育厅与思科(中国)创新科技有限公司产学合作协同育人项目(粤教高函[2017]153号)
本文URL http://www.arocmag.com/article/01-2020-06-013.html
英文标题 Clustering algorithm of social network based on graph clustering and ant colony optimization algorithm
作者英文名 Ye Xiaoying, Wan Mei, Tang Rong, Xie Yun, Chen Guihong, Li Qiang
机构英文名 1.Dept. of Computer Science & Technology,Neusoft Institute of Guangdong,Foshan Guangdong 528225,China;2.Dept. of Computer Science & Engineering,Guangzhou College of Technology & Business,Guangzhou 510850,China;3.Jiulongpo District Mental Health Center,Chongqing 400052,China;4.School of Electronics & Information Technology,Sun Yat-sen University,Guangzhou 510006,China
英文摘要 Aiming at the properties of direction and diversity of social relationships in the social network, this paper proposed a clustering algorithm of social network based on graph clustering and ant colony optimization algorithm. Firstly, it constructed a directed and non-fully connected complete graph for the social networks under constraint condition of network coverage; Then, it adopted K-medoids algorithm to search the center users of all user groups, adopted ant colony optimization to search the similarities of each user and center users in the graph, and grouped the users satisfied the threshold condition into the same group. This paper also designed a prediction mechanism of low active degree users to resolve the sparsity problem and cold-start problem. Besides, it set the network coverage constraint condition to balance the indexes of accuracy and coverage. Simulation experimental results indicate that the proposed algorithm realizes a good clustering performance of social network, and it reduces the problems of sparsity and cold-start effectively.
英文关键词 social network; data mining; clustering process; ant colony optimization; graph clustering; trust information
参考文献 查看稿件参考文献
 
收稿日期 2018/12/12
修回日期 2019/1/25
页码 1670-1674,1687
中图分类号 TP393
文献标志码 A