A Context Aware Recommendation System for movies

Abstract

People like to watch movies, but they do not always know what they want to see. Although there exist applications that helps people solve this issue, they are not being used very often (see Appendix A). People tend to do their own research to decide what to wach. This thesis will be looking into this issue and provide a proof of concept prototype. Not only does the prototype recommend movies, but it also introduces an social feature that can be extended to so much more. This feature were found by doing a round of interviews, while trying to nd the cause of why people do not use the recommendation systems. The thesis describes some of the state of the art approaches to recommenda- tion system algorithms and the implementation of the prototype. The prototype was evaluated by its accuracy and speed using data sets from Movielens. The results show that the chosen approach to recommendation systems is viable, and that the accuracy of the recommendation is not the only reason that recommendation systems are not being used that much. The results show that a social feature might increase the use of recommendation systems.