Leveraging recommendation systems for improving caching emerging short video in content delivery network

• DOI: 10.1002/ett.4117
• Publication Date: 2020-09-21T19:40:36Z
• Authors: Shuangyuan Li
• Keywords: Feature (linguistics); Content Delivery

Abstract Short video applications have rapidly become one of the most popular online entertainments. It is also important ways social networking. However, request pattern short videos shows characteristics highly dynamic compared to traditional video. First, access changing quickly, that is, user less preferences. Second, no tag descriptions. These result in a significant reduction caching efficiency content delivery network (CDN) server, which used cache videos. To address these challenges, this article proposes recommendation system optimize caching. For lack description, event detection method recognize tags based on improved dense trajectory and concept dictionary feature extraction. Then collaborative filtering generate list for CDN guide Simulation results show proposed can significantly improve CDN.