Popularity Prediction of Video Content Over Cloud-Based CDN Using End User Interest

Abstract

It is often believed that more is better, but that is not true in the case of data. As online data is increasing briskly, we are not able to handle such enormous data. With the increasing trends of speedy and uninterrupted access to data usage, CDNs have become quite popular in today’s world. However, it has become difficult to store all the content on CDN servers. This paper aims towards optimizing one of the aspects of CDN’s cached data that is video content. We propose a push-based caching approach by finding appropriate popular videos in accordance with a region to improve an end user’s quality of experience. A semi-supervised machine learning approach has been implemented to classify videos as low, medium, or highly popular. Popularity Prediction research has increased in energy lately. In any case, there has been little work done dependent on prior and significant video parameters for popularity prediction purposes. The experimental results show good accuracy, justifying the selection of parameters and the processing associated with them