一种新的复杂网络聚类算法  

New complex network clustering algorithm

作　　者：李峻金[1] 向阳[1] 牛鹏[1] 刘丽明[2] 芦英明 

机构地区：[1]西安通信学院 [2]中国人民解放军72556部队 [3]中国特种车辆研究所

出　　处：《计算机应用研究》2010年第6期2097-2099,共3页Application Research of Computers

摘　　要：揭示网络簇结构的复杂网络聚类方法研究具有重要的理论意义和应用价值。应用两种谱方法将复杂网络簇结构发现问题转换为空间数据聚类问题,并将粒子群聚类算法应用到对复杂网络簇结构的探测,提出了两种新的结合粒子群聚类的复杂网络簇结构探测算法。最后在两类复杂网络上进行实验并对实验结果进行了比较分析,提出的新算法在聚类准确性方面效果更好。Network clustering algorithms which aim to discover all natural network communities from given complex networks are fundamentally important for both theoretical researches and practical applications. This paper used two spectral partition methods in order to transform the communities detecting into cluster analysis problem. Then, applied PSO clustering algorithms to detect cluster structure. Proposed two new network clustering algorithms closely combined with PSO and demonstrated the availability of the algorithm in two different kinds of network datum. It also makes the comparison and analysis of the experimental results and obtains a conclusion that the proposed algorithms present fitness in clustering veracity.

关 键 词：复杂网络 网络聚类 网络簇结构 谱方法 粒子群聚类算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]