在线电影点播中的人类动力学模式  

Human Activity Pattern on On-line Movie Watching

作　　者：周涛[1] 

机构地区：[1]中国科学技术大学近代物理系

出　　处：《复杂系统与复杂性科学》2008年第1期1-5,共5页Complex Systems and Complexity Science

基　　金：973项目(2006CB705500);国家自然基金重点项目(10635040);国家自然基金面上项目(10472116);国家自然基金委主任基金(60744003);中国科学院院长特别基金

摘　　要：对人类行为的定量分析是理解很多社会经济复杂系统的基础。基于一个在线电影共享的大型数据库,给出了两次观看电影间隔时间分布的系统实证研究。研究发现在群体水平上,间隔时间分布可以用幂律函数近似刻画,并且,其幂指数和对应人群观看电影的活跃程度之间存在单调的关系。在个体层次上,观察到了具有胖尾特征的分布,以及分布宽度和活跃程度之间的负相关性。这些发现暗示了活跃程度在决定人类行为模式时的重要地位。活跃程度和幂指数之间的单调关系可能是一大类兴趣驱动模型的共同特征,这与以前在任务驱动系统中观察到的普适类有本质上的区别。The quantitative understanding of human behavior provides the elementary comprehension of the complexity phenomena of many human-initiated systems. In this paper, based on a huge database of an on-line movie-sharing system, we present a systematic empirical exploration on the distribution of interevent time between two consecutive watchings of movies. A monotonous relation between the activity and power-law exponent in the group level is found, and in the individual level, we observe heavy-tailed distribution for each user as well as negative correlation between activity and the width of distribution. These findings demonstrate the significant role of activity in determining the pattern of human behaviors, highlight a common character in the interest-driven human dynamics, and are different from the observed universality classes in the task-driven systems.

关 键 词：人类动力学 在线电影 时间间隔分布 活跃程度 

分 类 号：N94[自然科学总论—系统科学]