An entertainment recommendation system using the dynamics of user behavior over time

Personalized entertainment items recommendation is required to help millions of people narrow the universe of potential items to fit their unique tastes. These services usually depend on a machine-learning algorithm, which breaks down items into long lists of attributes and matches these elements to a user´s preferences. A set of such algorithms have been proposed. Most of these, such as collaborative filtering, works by finding a group of users based on the item he/she buys or provides feedback and then recommend popular items in the group. In this paper we argue that relying only on such acts are not sufficient for an effective recommendation system, we need to consider other things, such as, an entertainment item enjoying time since it could change over the course time. In this paper we proposed a novel algorithm to find a group of user that usage not only the ratings but also the time of the given ratings. Additionally, we propose algorithms for recommending items to the producers such that they can entertain us more. We perform experiments to validate the performance of our system. We show that our system outperforms in comparison with the existing algorithms.