基于度中心性局部扩展的社区划分算法  

Community Detection Algorithm Based on Degree-Centralized Local Extension

作　　者：柳曾雄 施化吉[1] 李雷[1] 施磊磊[1] 孙祥瑜 LIU Zengxiong;SHI Huaji;LI Lei;SHI Leilei;SUN Xiangyu(School of Computer Science&Communication Engineering,Jiangsu University,Zhenjiang 212013)

机构地区：[1]江苏大学计算机科学与通信工程学院,镇江212013

出　　处：《计算机与数字工程》2021年第10期2073-2077,2160,共6页Computer & Digital Engineering

摘　　要：社交网络中社区划分问题的研究不仅为网络演化、信息传播和影响力分析等方向提供了理论依据,而且在好友推荐、商业营销和舆情检测等领域有着重要应用价值。针对基于贪婪优化的社区划分算法AGSO不稳定问题,提出了一种基于度中心性局部扩展的社区划分算法(DCLE)。首先计算所有节点的度中心性(Degree Centrality),其次将链接两端节点度中心性之和作为链接的度中心性并降序排序,其后将度中心性最大链接作为初始链接加入网络,最后基于贪婪策略局部扩展并迭代,得到最终的社区划分结果。通过在公开的数据集和大型人工网络上进行实验,结果表明DCLE算法能快速且准确地发掘社区结构,稳定性得到显著提升。The research on community detection in social network not only provides theoretical basis for network evolution,information dissemination and impact analysis,but also has important application value in the fields of friend recommendation,commercial marketing and public opinion detection.Aiming at the instability problem of the greedy-optimization based community algorithm(AGSO),a degree-centralized local extension based community detection algorithm(DCLE)is proposed.Firstly,the degree centrality of all nodes is calculated.Secondly,the sum of the centrality of the nodes at both ends of the link is taken as the degree of the link and descended,and then the maximum link of the centrality as the initial link is added to the network.Finally,based on the greedy strategy,local expansion and iteration are used to obtain the final community division result.Experiments on public datasets and large-scale artificial networks show that the DCLE algorithm can quickly and accurately explore the community structure and the stability is significantly improved.

关 键 词：度中心性 社交网络 社区划分 局部扩展 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]