基于多目标优化的网络社区发现方法

发布时间：2019-05-30 04:02

【摘要】：社区发现是复杂网络挖掘中的重要任务之一,在恐怖组织识别、蛋白质功能预测、舆情分析等方面具有重要的理论和应用价值.但是,现有的社区质量评判指标具有数据依赖性与耦合关联性,而且基于单一评判指标优化的网络社区发现算法有很大的局限性.针对这些问题,将网络社区发现问题形式化为多目标优化问题,提出了一种基于多目标粒子群优化的网络社区发现算法MOCD-PSO,它选取模块度Q、最小最大割MinMaxCut与轮廓(silhouette)这3个指标进行综合寻优.实验结果表明,MOCD-PSO算法具有较好的收敛性,能够发现分布均匀且分散度较高的Pareto最优网络社区结构集,并且无论与单目标优化方法(GN与GA-Net)相比较,还是与多目标优化算法(MOGANet与SCAH-MOHSA)相比较,MOCD-PSO算法都能在无先验信息的条件下挖掘出更高质量的网络社区.
[Abstract]:Community discovery is one of the important tasks in complex network mining, which has important theoretical and application value in terrorist organization identification, protein function prediction, public opinion analysis and so on. However, the existing community quality evaluation indicators have data dependence and coupling correlation, and the network community discovery algorithm based on single evaluation index optimization has great limitations. In order to solve these problems, the network community discovery problem is transformed into a multi-objective optimization problem, and a network community discovery algorithm based on multi-objective particle swarm optimization (MOCD-PSO,) is proposed, which selects modularity Q. The three indexes of minimum maximum cut MinMaxCut and outline (silhouette) are optimized synthetically. The experimental results show that the MOCD-PSO algorithm has better convergence and can find the Pareto optimal network community structure set with uniform distribution and high dispersion, and compared with the single objective optimization method (GN and GA-Net). Compared with multi-objective optimization algorithm (MOGANet and SCAH-MOHSA), MOCD-PSO algorithm can mine higher quality network community without prior information.
【作者单位】：福建师范大学软件学院;广西师范大学计算机科学与信息工程学院;School
【基金】：国家自然科学基金(61170131,61263035) 澳大利亚ARC(DP0985456) 国家高技术研究发展计划(863)(2012AA011005) 国家重点基础研究发展计划(973)(2013CB329404) 教育部人文社会科学研究青年基金(12YJCZH074) 福建师范大学优秀青年骨干教师培养基金(fjsdjk2012082) 科学计算与智能信息处理广西高校重点实验室开放基金(GXSCIIP201212)
【分类号】：TP18;TP393.09

【相似文献】