This research was carried out as a part of 99X Technology’s project “Televisionary”. In this we implemented a system which is capable of recommending TV shows for TV fans based their interests. The system is based on collaborative filtering which is a modern Artificial Intelligence technique. The method concerns the user’s perspective of a particular TV show. By using collaborative filtering it can predict the rating value (0;5 points) for a given user for a specific TV show. Subsequently these predicted ratings are used to guess the TV shows that best fit for the selected user. Information such as user, movie and ratings are typically used in collaborative filtering algorithms to determine the recommendations for a user. Going beyond this traditional model we introduced a fourth dimension, tags which gives a clear idea about the user perspective related to the TV show. These tags describe why that particular user likes or keep track on that given TV show. Since the ratings cannot be considered as a clear indication of the user’s view towards the movie this fourth dimension helps to add semantics so that the predictions made are more accurate. These semantics are used to perform semantically enhanced collaborative filtering and make better recommendations for the users.

Collaborated with: Faculty of Information Technology, University of Moratuwa