面向复杂网络的Louvain改进算法研究与探讨
Research and discussion on improved louvain algorithm for complex network
徐宏哲 1莫欣岳 1李欢1
作者信息
摘要
[目的/意义]社团检测作为复杂网络分析的一个主要研究方向,能够探究复杂网络的社团结构特性,挖掘潜藏在背后的关联信息,可用于网络安全治理,识别和划分潜在的危险分子,维持网络的安全现状.[方法/过程]在Louvain算法步骤的合并过程中,存在着合并社团时选择局部最优解而并非全局最优解的问题.为了缓解这个问题,使用K-means的中心初始化思路改进,提出了基于节点相异度初始化社团的Louvain改进算法(简称LBNDC算法).[结果/结论]经过仿真实验和经典算法对比,LBNDC算法表现优秀,为网络安全中复杂网络的社团检测提供了新的算法设计思路.
Abstract
[Purpose/Significance]As a major research direction of Complex network analysis,community detection can explore the community structure characteristics of Complex network,mine the underlying associated information,and be used for the governance of network security,identify and classify potentially dangerous molecules,and maintain the status quo of security.[Method/Process]In the merging process of the Louvain algorithm steps,there is a problem of selecting local optimal solutions instead of global optimal solutions when merging communities.To alleviate this problem,the central initialization idea of K-means is used to improve the Louvain improved algorithm based on node dissimilarity initialization of communities,abbreviated as LBNDC algorithm.[Results/Conclusion]After simulation experiments and comparison with classic algorithms,the LBNDC algorithm performs well.It provides a new algorithm design idea for community detection of Complex network in network security.
关键词
复杂网络/社团检测/Louvain算法/节点相异度/网络安全Key words
complex networks/community detection/louvain algorithm/node dissimilarity/network security引用本文复制引用
基金项目
教育部产学合作协同育人项目(220902070162538)
中国高等教育学会高等教育科学研究规划课题(22LH0409)
海南省自然科学基金(623RC455)
海南省自然科学基金(623RC457)
海南大学科研启动基金项目(KYQDZR-22096)
海南大学科研启动基金项目(KYQDZR-22097)
海南大学教育教学改革研究项目(hdjy2364)
出版年
2024
NETL
NSTL
万方数据