Characterization and Modeling of Video Popularity
This paper presents a detailed characterization of video popularity for a video-on-demand system at University of Science and Technology- of China. The paper proposes to quantitatively characterize the video popularity with three factors, including the common access frequency, the ratio between video playback duration and video length and jump backwards frequency. Numerical studies on a real workload show that the video popularity distribution is better modeled with the stretched exponential distribution than Zipf-like distribution. It is also shown that the shape of the stretched exponential distribution of popularity- is determined by both the duration of workload and the characterizing method. Moreover, it analyzes the relationship between the importance of the access frequency and that of other two factors, and finds the discrimination of popularity characterized only by the access frequency could be significantly improved by simultaneously considering other two factors.
video popularity stretched exponential distribution zipf-like distribution video-on-demand system
Lei Chen Song Wang You Zhou Gang Wu Tiebing Xie
Key Laboratory of Network Communication System and Control,CAS University of Science and Technology Engineering Research Center of Networking & Digital Signal Processing Institute of Acoustics,Chinese
国际会议
太原
英文
159-164
2011-02-26（万方平台首次上网日期，不代表论文的发表时间）