摘要
分析了基于优化模块度检测复杂网络社区结构的算法存在解的限制问题,即不能检测出小于一定内在尺度的社区,并提出了基于极值优化模块密度来检测复杂网络社区结构的启发式算法,通过调整局部极值来优化全局的变量,使算法具有更好的持续搜索和跳出局优解的能力.通过人工网络和现实网络实验分析表明,本文算法用于检测大型网络社区时,具有较高的正确率和效率,即使当网络结构变得很模糊时,算法也能很好地工作.
Taking modularity as an objective function, many algorithms for detecting community structure in complex networks were proposed, which failed to identify modules smaller than an intrinsic scale. After the resolution limits in community detection were studied, a heuristic algorithm based on extremal optimization modularity density was designed which operates optimizing a global variable by improving extremal local variables and has better ability of continuing search and jumping out of local optimal solution. Simulation tests on artificial networks and real networks show that the algorithm has good performances in detecting large networks community structure. Particularly, when community structure is obscure in the cases, the algorithm can also work well
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2011年第4期82-85,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60873099)
关键词
复杂网络
聚类算法
启发式算法
社区检测
极值优化
模块密度
complex networks
clustering algorithms
heuristic algorithms
community detection
extremal optimization
modularity density
