基于联合矩阵分解的动态异质网络社区发现方法

发布时间：2018-04-20 16:40

本文选题：异质网络 + 动态网络　；参考：《计算机应用研究》2017年10期

【摘要】：动态网络的社区发现是目前复杂网络分析领域的重要研究内容,然而现有动态网络社区发现方法主要针对同质网络,当网络包含多种异质信息时,现有方法不再适用。针对这个问题,提出了一种基于联合矩阵分解的动态异质网络社区发现方法。首先计算动态异质网络中各个快照图的拓扑相似度矩阵和多关系相似度矩阵;其次利用时序联合非负矩阵分解方法,约束各个时刻快照图的社区划分;最后在真实网络数据集上与K-means、Meta Fac算法进行比较实验,提出算法能够充分利用网络的异质信息与拓扑信息,异质网络社区划分精度优于Meta Fac算法,且划分效果更稳定。结果表明,基于联合矩阵分解的动态异质网络社区发现算法可以有效检测出动态异质网络中潜在的社区结构。
[Abstract]:Community discovery of dynamic networks is an important research content in the field of complex network analysis. However, the existing dynamic network community discovery methods are mainly aimed at homogeneous networks. When the network contains a variety of heterogeneous information, the existing methods are no longer applicable. To solve this problem, a dynamic heterogeneous network community discovery method based on joint matrix decomposition is proposed. First, the topological similarity matrix and multi-relation similarity matrix of each snapshot graph in dynamic heterogeneous network are calculated; secondly, the community partition of snapshot graph at each moment is constrained by using the method of time-series combined with non-negative matrix decomposition. Finally, by comparing the real network data set with the K-means-Meta Fac algorithm, it is proposed that the algorithm can make full use of the heterogeneous information and topology information of the network, and the accuracy of the heterogeneous network community division is better than that of the Meta Fac algorithm, and the partition effect is more stable. The results show that the community discovery algorithm based on joint matrix decomposition can effectively detect the potential community structures in dynamic heterogeneous networks.
【作者单位】：国家数字交换系统工程技术研究中心;
【基金】：国家科技支撑计划资助项目(2014BAH30B01) 国家自然科学基金创新群体资助项目(61521003);国家自然科学基金资助项目(61379151)
【分类号】：O157.5;TP301.6

【相似文献】