A novel approach to predict the popularity of the video

Abstract

Video classification is an emerging research area. Advertisement videos play a major role in improving targeted sales of a product. The popularity of the advertisement depends on the number of viewers attracted to that advertisement video in a short span of time. Popular advertisements have more impact on the sales of the product. In this paper, we propose a new measure called Video Differential Index(VDI) to analyze and classify advertisement videos to know whether that video attracts more number of viewers or not. By using VDI measure we can also predict the type of video i.e., action-oriented, theme or concept based without watching the content of the video. Our experimental results demonstrated that, action oriented videos have high Video Differential Index as compared to theme or concept based videos. The VDI can be directly related to the popularity of the video.