一种基于概率模型的加权网络社群结构探测方法 a probability model .pdf

一种基于概率模型的加权网络社群结构探测方法 1,2 2* 张锴琦 杜海峰 (1. 西安交通大学 管理学院，陕西 西安 710049 ； 2.西安交通大学公共管理与复杂性科学研究中心，陕西 西安 710049) 摘 要 加权网络的社群结构探测是社群结构探测问题的难点和热点。Brian 等提出的基于概率模型的重叠社群探测方法（Principled statistical approach for overlapping communities，简称PSOC 算法）在解决重叠社群结构探测问题的同 时，为加权网络的社群结构探测提供了良好的研究基础。本文在PSOC 的算法的 设计模型基础上，结合已有对于加权网络的研究提出了一种针对加权网络的社群 结构探测算法。通过对标准测试网络的社群结构探测实验，对所提算法的有效性 和可用性进行了检验。结果表明，本文所提算法能够较好的对加权网络中社群结 构进行探测，且富有效率。并且，本文对实际加权网络的社群结构进行了探测， 对其探测结果及社会学意义进行了讨论。 关键词 社群结构 加权网络 概率模型 算法 A probability model algorithm for weighted network community structure detection Zhang Kaiqi1,2 Du Haifeng2 (1. School of Management, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710049, China； 2. Center for Administration and Complexity Science, School of Public Policy and Administration, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710049, China;) ABSTRACT Weighted network community structure detection is difficult and hot issues in community structure detection. The principled statistical approach for overlapping communities (PSOC approach) proposed by Brian et al. provides a good study basis for weighted network community structure detection. In this paper we propose a probability model algorithm for weighted network community structure detection based on PSOC approach and other relative research. And we test the effectiveness and availability of the proposed algorithm based on the community structure detection experiment on standard test network. The results show that the proposed algorithm can detect the weighted network community structure efficiently. Finally we detect the real weighted network community structure and discuss the sociological sense of the detection results. KEYWORD Community structure; Weighted network; Probabi