A Time-context-Based Collaborative Filtering Algorithm

Collaborative filtering, one of the most widely used algorithm in recommender system, predicts a user's preference towards an item by aggregating ratings given by users having similar taste with that user. State-of-the-art approaches introduce many other secondary methods to combine to cope with sparsity and precision problem. However, these hybrid approaches rarely consider the importance of context information. This paper incorporates the time-context, one of the most important contexts, into the traditional collaborative filtering algorithm and proposes a time-context-based collaborative filtering (TBCF) algorithm to improve the performance for traditional collaborative filtering algorithm. Experiments evaluating our approach are carried out on real dataset taken from movie recommendation system provided by MovieLens Web site. The result shows the proposed approach can improve predication accuracy and recall ratio compared with existing methods.