重大事件下微博用户行为统计分析及建模

基于实证的人类动力学研究认为人类活动的时间间隔分布具有胖尾特性。通过研究新浪微博——中国最成功的在线社交系统上所搜集到的数据，选取四类情况（包括三类重大事件以及普通情况），分析微博用户发布微博、转发微博以及评论微博的情况。研究发现四类情况下微博发布、转发和评论的时间间隔分布均服从幂律分布，并因为事件影响时长而互有差别。在微博用户行为统计分析的基础上，借鉴排队论和自适应网络提出了一个兴趣驱动模型，该模型能够很好地解释幂律分布与各事件影响时长之间的关系，这些发现对于探索信息传播机制的规律有一定帮助。

The Behavior Analysis and Modeling of Micro-Blog users under specific events

Empirical evidences indicate that the inter-event time distribution of human behavior exhibits heavy- tailed features. This paper studied the interest-driven human dynamics using data from Sina micro-blog, which is one of the most successful on-line social media systems in China. It chose three specific events and a period of time called ＂general situation＂ in which no big events occur. The results of data analysis showed that the inter- post, inter-comment and inter-forward time distributions of each of these events follow the power law distribution. Then this paper proposed an interest-driven model with the concept of batch customer arrival in the Queuing The- ory. And this model well explains the relationship between the power-law exponent and the effective duration of each event. The findings are helpful for discovering regular patterns of the message dissemination.