Understanding Video Propagation in Online Social Networks: Measurement, Analysis, and Enhancement

The deep penetration of Online Social Networks (OSNs) has made them major portals for video content sharing recently. It is known that a significant portion of the accesses to video sharing sites (VSSes) are now coming from OSN users. For example, YouTube reported that, as of January 2012, more than 700 tweets per minute containing a YouTube link, and over 500 years' worth of YouTube videos are watched by Facebook users every day. Although the videos shared in OSNs are mostly from VSSes, OSNs provide quite different mouth-to-mouth-like sharing mechanisms, leading to distinctive user access patterns. Yet the unique features of video sharing over OSNs and their impact remain largely unknown. In this thesis, we conduct a systematic study on the video propagation in OSNs based on large-scale real-world data. Our study unveils the unique characteristics of video requests from OSNs, showing that an OSN can dramatically amplify the skewness of video popularity that 2% most popular videos account for 90% of total views; and video popularity also exhibits much more dynamics with multiple request bursts. We then closely analyze the video propagation process in OSNs with both measurement and modeling, identifying the key influential factors. We further examine the popularity prediction of videos shared in OSNs. We demonstrate that conventional methods largely fail in this new context, and develop a novel propagation-based prediction model. Finally, based on the above studies, we present SNACS (Social Network Aware Cloud Assistance for Video Sharing), which enables OSN operators to cost-effectively enhance the video viewing experience of their users through utilizing content cloud services.