视频点播中视频流行度的建模与分析

Modeling and analysis of video popularity in video-on-demand

通过对视频点播中的用户行为进行研究,可以改善系统性能和用户观看体验。视频流行度是视频点播系统用户行为的主要特征,本文采用广延指数模型和漂移幂律模型对其进行建模,通过对比分析,结果表明短期内,漂移幂律模型更好的刻画了视频流行度;长期内,广延指数模型更好的刻画了视频流行度。

Through the study of user behavior in video-on-demand, it can improve the system performance and user experience. Video popularity is the main feature of user behavior in video-on-demand system. In this paper, the extended exponential model and the drift power law model are used to model the distribution of video popularity. Through comparative analysis, the result show that in the short-term, the drift power law model better characterizes the video popularity; in the long-term, the extended exponential model better characterizes the video popularity.