基于主题和链接分析的微博社区发现算法

针对传统社区发现方法大多基于链接或主题关系, 且没有考虑获取微博用户社会信息时的限制, 无法有效识别微博中多个社区的问题, 提出了一种综合基于主题和链接分析的微博社区发现算法来挖掘微博中多个社区。算法首先研究微博用户的链接及博文主题特性, 定义了链接相关度和主题相关度公式; 然后推出用户总相关度公式, 以此来计算节点间的传递概率, 用改进后的标签传递算法对用户分类; 最终划分出兴趣相似且社会联系紧密的用户群。真实数据集上的仿真实验验证了该方法的合理性和有效性。

Community discovery for microblog based on topic and link analysis

Tranditional community discovery algorithms are generally based on either links or interests and don't take limits of obtaining microblog users' social information into consideration, so they can't detect multiple communities effectively. Therefore, this paper proposed a microblog community discovery algorithm based on both links and topics to discover communities in microblog. It first studied characteristics of links and blog's topics, then deduced user's relationship formulas, on basis of which, it calculated transfer probability and used improved label propagation algoritym to divide communities. Finally, it distinguished different clusters of people who close relationships and similar interests. The simulation results on real social dataset verify that the proposed method is reasonable and effective.