层次聚类社区发现算法的研究

Survey on algorithms of community detection

概述了社区发现算法的研究现状;介绍了因分析对象的不同而产生的四类社区发现方法:矩阵谱分析方法、层次聚类方法、基于边图思想的方法和基于极大团思想的方法。对其中性能最优的层次聚类方法进行了详细的综述,并对其典型算法进行了分析比较。最后,提出了社区发现算法可能的研究方向,为今后的研究提供参考。

This paper briefly introduced research situation of community detection. Secondly, it classified community detection methods into broad categories as spectral algorithms, hierarchical clustering, link clustering and clique percolation, then gave a brief introduction about them. Since hierarchical clustering had better performance, it especially presented and compared some typical methods of it. Finally, it pointed out the future research directions that would be hopefully beneficial to the researchers from related fields.