论文部分内容阅读
微博中的意见领袖对信息的快速传播起着关键作用,能在短时间内对数量众多的用户产生直接或间接的影响.在微博中,意见领袖除了具有自身属性和网络结构特征外,还与参与的话题高度相关.针对已有挖掘研究只考虑了意见领袖的局部特征以及忽略了话题相关性的问题,提出一种话题相关的意见领袖挖掘算法.该算法首先根据微博用户的自身属性及用户间话题相关的交互信息构建带权的话题相关的微博图模型,并采用随机游走的思想来寻找图模型的中心节点,以此挖掘微博中的意见领袖.在新浪微博三个话题数据集上的实验结果表明,该算法挖掘的意见领袖在扩展核心率指标上优于类似算法.
Opinion leaders in Weibo play a key role in the rapid spread of information and can have a direct or indirect impact on a large number of users in a short period of time.In addition to their own attributes and network structure, But also has a high correlation with the topic of participation.According to the existing mining research which considers only the local characteristics of opinion leaders and ignores the topic relevance, this paper proposes a topic-related opinion leader mining algorithm.This algorithm first based on the Weibo user’s own Attributes and inter-user topics related to interactive information to build a weighted topic-related microblogging graph model, and use the random walk to find the center node of the graph model in order to tap opinion leaders in the microblogging in Sina microblogging The experimental results on the three topic datasets show that the opinion leaders proposed by this algorithm are superior to similar algorithms in extending the core rate index.