基于多核心节点的增量式动态社区发现算法  

Incremental dynamic community discovery algorithm based on multi-core nodes

作　　者：陈晶 刘志君 杨新宇 刘洺辛 刘苗苗 CHEN Jing;LIU Zhijun;YANG Xinyu;LIU Mingxin;LIU Miaomiao(College of Mathematics and Computer Science,Guangdong Ocean University,Zhanjiang 524088,China;College of Information Science and Engineering,Yanshan University,Qinhuangdao 066004,China;Key Laboratory of Virtual Technology and System Integration,Qinhuangdao 066004,China;College of Electronic and Information Engineering,Guangdong Ocean University,Zhanjiang 524088,China;School of Computer&Information Technology,Northeast Petroleum University,Daqing 163318,China)

机构地区：[1]广东海洋大学数学与计算机学院,广东湛江524088 [2]燕山大学信息科学与工程学院计算机系,河北秦皇岛066004 [3]河北省虚拟技术与系统集成重点实验室,河北秦皇岛066004 [4]广东海洋大学电子信息工程学院,广东湛江524088 [5]东北石油大学计算机与信息技术学院,黑龙江大庆163318

出　　处：《通信学报》2024年第3期66-80,共15页Journal on Communications

基　　金：国家自然科学基金资助项目(No.62172352,No.42306218);中央省部共建基金资助项目(No.226Z0102G,No.226Z0305G);河北省自然科学基金资助项目(No.2022203028,No.F2023407003);广东海洋大学科研启动基金资助项目(No.060302102304)。

摘　　要：针对动态社区发现算法通常基于社区结构平稳变化的假设,而难以应对演化过程中可能出现的大量社区消亡或涌现等突发事件的问题,提出了一种基于多核心节点的增量式动态社区发现算法MCNIDCD。首先,将核心节点分为扩散型和内聚型,制定4种增量更新策略。其次,通过局部更新调整节点社区归属,并采用增量模块度方法优化社区结构。最后,实现社区合并。在人工和真实网络上对该算法的性能进行了评估,实验结果表明,在对比目前相关动态社区检测算法时,在人工网络仿真环境中,MCNIDCD算法表现出与社区演化规律的高度契合性;在真实网络实验中,MCNIDCD算法在模块度性能指标上平均提升了28%,并且在稳定性方面具有良好的优势,其优势对于研究动态社区演化过程具有重要的意义。A new incremental dynamic community discovery algorithm MCNIDCD based on multiple core nodes was proposed to address challenges in dynamic community discovery.It adapted to sudden events like the emergence or disappearance of communities during evolution.MCNIDCD categorized core nodes into diffusion and cohesion types,and devised four incremental updating strategies.It adjusted node community membership locally and optimized community structure using an incremental modularity method to facilitate community merging.Evaluation on artificial and real networks shows MCNIDCD’s high conformity to community evolution patterns.In real network experiments,MCNIDCD exhibits a 28%average improvement in modularity performance and significant stability advantages.Its superiority is important for studying dynamic community evolution.

关 键 词：核心节点 增量式 网络演化 社区发现 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]