Temporal Recommendation System Based on Coupled Tensor Factorization

Recommender systems analyze the user’s preference patterns and provide personalized recommendations of items that will suit a user’s taste. An essential challenge in these systems is that user preferences are not static and users are likely to change their preferences over time. The adaptability of recommender systems to capture the evolving user′s preferences which are constantly changing, improves the accuracy of recommender systems. In this paper, we develop a model to capture the users’ preference dynamics in a personalized manner. We introduce an individual time decay factor for each user according to the rate of his preference dynamics to weight the past user preferences and decrease gradually the importance of them. We exploit the users’ demographics as well as the extracted similarities between users over time, in addition to the past weighted user preferences, in a developed coupled tensor-matrix factorization technique to provide the personalized recommendations. Our evaluation results on the two real-world datasets indicate that our proposed model is better and more robust than the competitive methods in term of recommendation accuracy and is more capable to cope with cold-start problem.