Video Recommendation System for YouTube Considering Users Feedback

Abstract

Youtube is the most video sharing and viewing platform in the world. As there are many people of different tastes, hundreds of categories of videos can be found on YouTube while thousands of videos of each. So, when the site recommends videos for a user it takes some issues which fill the needs of the user. Most of the time a user watches videos related to the previously watched video. But sometimes userFFFD;s mood changes with time or weather. A user may not hear a song in the whole year but can search the song on a rainy day. Another case a user may watch some types of videos at day but another type of videos at night or another at midnight. In this paper, we propose a recommendation system considering some attributes like weather, time, month to understand the dynamic mood of a user. Each attribute is assigned a weight calculated by performing a survey on some YouTube users. Most recently viewed videos is assigned heavy weight and weather is assigned lower. This recommendation system will make YouTube more user-friendly, dynamic and acceptable.