Author:

Keywords:

Science & Technology, Technology, Computer Science, Theory & Methods, Engineering, Electrical & Electronic, Computer Science, Engineering, User-generated content, YouTube, tag, prediction

Abstract:

© 2016 IEEE. A large portion of today's Internet traffic originatesfrom streaming and video services. Such services rely on acombination of distributed datacenters, powerful content deliverynetworks (CDN), and multi-level caching. In spite of this infras-tructure, storing, indexing, and serving these videos remains adaily engineering challenge that requires increasing efforts onthe part of providers and ISPs. In this paper, we explore howthe tags attached to videos by users could help improve thisinfrastructure, and lead to better performance on a global scale. Our analysis shows that tags can be interpreted as markers ofa video's geographic diffusion, with some tags strongly linkedto well identified geographic areas. Based on our findings, wedemonstrate the potential of tags to help predict distributionof a video's views, and present results suggesting that tags canhelp place videos in globally distributed datacenters. We show inparticular that even a simplistic approach based on tags can helppredict a minimum of 65.9% of a video's views for a majorityof videos, and that a simple tag-based placement strategy is ableto improve the hit rate of a distributed on-line video service byup to 6.8% globally over a naive random allocation.