基于交互行为和连接分析的社交网络社团检测
发布时间：2019-06-26 12:38
【摘要】：社交网络的迅猛发展极大地方便了人们的日常生活、工作和学习,但也带来了大量复杂的交互行为和连接模式。如何有效地综合分析网络中的交互信息和网络节点之间存在的连接信息,进而完成高效的社团检测,是在当前网络多维属性的复杂背景下进行网络分析所面临的关键难题。基于此,从有效融合两类不同的异质信息研究出发,提出了一种基于交互行为和连接分析的社交网络社团检测(CDUILS)方法。该方法基于两类信息能够从不同的角度反映网络同一个社团归属的假设,采用联合非负矩阵分解架构,以迭代更新的方式,同时利用两类信息进行社团结果的获取。在真实网络数据集上的实验表明,与已有方法相比,所提方法能够有效融合两类信息进行社团检测,取得了更好的社团划分质量。
[Abstract]:The rapid development of social networks greatly facilitates people's daily life, work and study, but also brings a large number of complex interactive behavior and connection patterns. How to effectively analyze the interactive information in the network and the connection information between the network nodes, and then complete the efficient community detection, is the key problem in the network analysis under the complex background of the multi-dimensional attributes of the current network. Based on this, a social network community detection (CDUILS) method based on interaction behavior and connection analysis is proposed based on the effective fusion of two different kinds of heterogeneous information. Based on the assumption that the two kinds of information can reflect the same community belonging to the network from different angles, the joint non-negative matrix decomposition architecture is adopted to update the community results iteratively, and the two kinds of information are used to obtain the community results at the same time. The experiments on the real network data set show that compared with the existing methods, the proposed method can effectively fuse the two kinds of information for community detection, and obtain better community division quality.
【作者单位】： 国家数字交换系统工程技术研究中心;
【基金】：国家自然科学基金创新群体项目(61521003) 国家重点基础研究发展计划资助项目(2012CB315901,2012CB315905) 国家科技支撑计划(2014BAH30B01)资助
【分类号】：TP393.09
[Abstract]:The rapid development of social networks greatly facilitates people's daily life, work and study, but also brings a large number of complex interactive behavior and connection patterns. How to effectively analyze the interactive information in the network and the connection information between the network nodes, and then complete the efficient community detection, is the key problem in the network analysis under the complex background of the multi-dimensional attributes of the current network. Based on this, a social network community detection (CDUILS) method based on interaction behavior and connection analysis is proposed based on the effective fusion of two different kinds of heterogeneous information. Based on the assumption that the two kinds of information can reflect the same community belonging to the network from different angles, the joint non-negative matrix decomposition architecture is adopted to update the community results iteratively, and the two kinds of information are used to obtain the community results at the same time. The experiments on the real network data set show that compared with the existing methods, the proposed method can effectively fuse the two kinds of information for community detection, and obtain better community division quality.
【作者单位】： 国家数字交换系统工程技术研究中心;
【基金】：国家自然科学基金创新群体项目(61521003) 国家重点基础研究发展计划资助项目(2012CB315901,2012CB315905) 国家科技支撑计划(2014BAH30B01)资助
【分类号】：TP393.09
【参考文献】
相关期刊论文 前2条
1 常振超;陈鸿昶;刘阳;于洪涛;黄瑞阳;;基于联合矩阵分解的节点多属性网络社团检测[J];物理学报;2015年21期
2 许为;林柏钢;林思娟;杨e，
本文编号：2506184
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2506184.html
