计算机科学与探索 ›› 2021, Vol. 15 ›› Issue (4): 612-630.DOI: 10.3778/j.issn.1673-9418.2008023

• 综述·探索 • 上一篇    下一篇

动态社区发现方法研究综述

端祥宇,袁冠,孟凡荣   

  1. 1. 中国矿业大学 计算机科学与技术学院,江苏 徐州 221116
    2. 矿山数字化教育部工程研究中心,江苏 徐州 221116
  • 出版日期:2021-04-01 发布日期:2021-04-02

Dynamic Community Detection: A Survey

DUAN Xiangyu, YUAN Guan, MENG Fanrong   

  1. 1. School of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China 
    2. Digitization of Mine, Engineering Research Center of Ministry of Education, Xuzhou, Jiangsu 221116, China
  • Online:2021-04-01 Published:2021-04-02

摘要:

随着社交媒体多样性的增加,实时分析社交网络的需求不断增大,动态社区发现的研究受到了广泛的关注。已有的社区发现综述多是侧重静态社区发现,以及相关方法的探讨,无法进行网络演化分析,此外社区的实体数据往往具有交叉更替性和时序性,因此对动态社区发现的研究现状进行分析和综述。首先,基于复杂网络的研究背景,提出了通用的动态社区发现研究框架;接着,形式化表示动态社区发现的相关定义,并从网络层面和节点层面对动态社区演化进行详细分析;然后,根据架构和技术的不同,对动态社区发现方法进行归纳分类,并结合常用数据集和评价指标对经典静态社区发现算法进行定性和定量分析;最后,介绍了社区发现的典型应用场景,探讨了当前动态社区发现研究面临的主要挑战,针对性地提出了相关解决方案,为动态社区发现研究领域勾画出较为清晰和全面的研究方向。

关键词: 动态社区发现, 社交网络, 网络分析, 动态社区演化

Abstract:

With the increasing diversity of social media, the demand for real-time analysis of social networks continues to increase, and research on dynamic community detection has received extensive attention. Existing community detection reviews mostly focus on static community detection and the discussion of related methods, so network evolution analysis cannot be carried out. Besides, the entity data in the community are cross-substitutional and sequential. Therefore, the research status of dynamic community detection is studied, reviewed, and analyzed. First, based on the research background of complex networks, a general research framework for dynamic community detection is proposed. Then, the relevant definitions of dynamic community detection are formally expressed, and the evolution of dynamic communities at the network level and node level is analyzed in detail. According to the difference in architecture and technology, the dynamic community detection methods are summarized and classified. Combined with commonly used data sets, the static community detection algorithm is analyzed qualitatively and quantitatively with evaluation criteria. Finally, the typical application scenarios of community detection are introduced, the main challenges faced by current dynamic community detection research are discussed, and relevant solutions are put forward for dynamic communities, which outlines a clearer and comprehensive research direction for dynamic community detection research field.

Key words: dynamic community detection, social network, network analysis, dynamic community evolution