User preference and behavior pattern in Push VOD systems

Abstract

Push-VOD (Video on Demand) is a technique preloading the right video contents to end users used by TV suppliers. This system would benefit from knowing when and what a user watches. Thus there is a need for analyzing and predicting user behaviors. In this paper, we study user model from two aspects, preference and behavior pattern. For the preference part, we study what users watch. We build up the TV user model based on vector spaces. In our model, vectors represent the user profiles and video features. We mainly analyze user watching records and rankings towards videos based on the data from movie-lens and give our conclusions. We also give our survey and research on user behavior pattern, towards when user watches, namely different hours in a day and different days in a week.