The constant time of predictive algorithm for music recommendation with time context

In this research, we propose using time context to improve predictive accuracy and quality of collaborative filtering for music recommendation. We use time contextual information called micro-profiling. Thus, each user has multiple micro profiles, in particular, six-time slots instead of a single profile. The recommendation is performed depended on these micro-profiling. Our method takes into account time intervals in listening music represent user preference in up-to-date. In rating prediction approach, we adopt tendencies-based algorithm, one of the collaborative filtering algorithms, which is very low computational time and memory requirements. From the experimental results, it has shown that the performance of our approach gives more accuracy and low computational time than traditional CF and CF with matrix factorization.