摘要： 分析了常见的社团发现算法的特点，以及谱二分法在实际应用中必须不断迭代才能完成多社团发现的不足，并提出了基于Laplace图谱和K-Means聚类算法的多社团发现方法，该方法是一个可视化的决策过程。根据Laplace图谱的次小特征值和第三小特征值对应的特征向量，构成聚类样本并显示出来。根据决策者的意图，由决策者来确定社团的个数和聚类中心，应用K-Means聚类算法一次完成多社团的分类。

关键词: 复杂网络, 社团结构, Laplace图谱, K-Means算法, 可视化

Abstract: The characteristics of common community finding algorithm and the drawback of spectral bisection method in application are analyzed. The method of multi-community finding in complex networks based on Laplace graph spectrum and K-Mean is provided. It is a visualized decision process to compose clustering sample with the eigenvectors of the second and third minimum eigenvalue of Laplace graph spectrum, and to display it for decision maker to define the number of community and the centers of cluster on his intention, and then to apply K-Mean algorithm to get
the cluster of community in one time.

Key words: complex networks, community structure, Laplace graph spectrum, K-Means, visualization

中图分类号: 

• TP18