论文部分内容阅读
微博话题传播的时效性十分重要。文章首先提出话题时变传播网络定义,网络中的节点由参与话题的用户构成,用户转发微博话题的关系抽象成带时间戳的边。之后将KDD2013新浪微博话题数据建立了相应的时变传播网络,分析其结构发现:网络上两节点之间的信息延迟集中在1~5小时之间,且信息延迟服从幂律分布;用户具有随时间变化的动态影响力,是以时间为自变量的负指数函数。其次提出基于用户时间影响力驱动的微博话题传播模型TIM,模型主要规则为一个用户是否传播微博话题依赖于其所有邻居的时间影响力之和以及自身的传播阈值。最后在新浪微博关注关系网络上仿真运行模型机制,分析了时变传播网络结构与多个模型参数之间的关系。
Timeliness of microblogging topic communication is very important. Firstly, the paper proposes the definition of time-varying communication network, the nodes in the network are composed of the users who participate in the topic, and the relationship between the users who forward the Weibo topic and the time-stamped edge is abstract. Then the KDD2013 Sina microblogging topic data to establish a corresponding time-varying propagation network analysis of its structure found that: the network between the two nodes of information delay concentrated between 1 to 5 hours, and the information obeys power-law distribution delay; the user has The dynamic influence over time is a negative exponential function with time as an independent variable. Secondly, we propose TIM based on users¡¯ time influence driven microblog topic propagation model. The main rule of the model is whether a user propagates the weibo topic depends on the sum of the temporal influence of all his neighbors and his own propagation threshold. Finally, we focus on the simulation operation model of the relationship network in Sina Weibo, and analyze the relationship between the time-varying propagation network structure and multiple model parameters.