一种局部最优社区挖掘方法

摘要

研究互联网论坛中划分用户社区问题.首先通过分析用户在论坛上的发言层次结构与内容建立用户之间的回复关系图,然后提出一种基于局部最优的图聚类方法LOGCA对大容量的论坛网络图进行分类.实验得到互联网论坛上几个有意义的用户社区,并且确定了社区成员的共同兴趣.实验结果表明新方法简单有效,能够从大规模网络中发掘出有意义的用户社区.%This paper studied how to mine meaningful Web communities from forum. Constructed user interaction graph according to the hierarchical structure and content of forum posts. Proposed a novel graph clustering algorithm LOGCA to exploit meaningful communities from large-scale networks. The experimental results show that this approach is effective in discovering Web communities from Internet forum.