A Hybrid Recommendation Algorithm using Tags, Time and User Relationship

Abstract

A user behavioral preference is often influenced by a variety of factors, such as user relationship, time and so on. If only consider a single factor, it’s hard to make accurate recommendations. Therefore, this paper proposes a hybrid recommendation algorithm that considers tag semantic, user relationship and the time factor. Firstly, modeling the user’s tagging behavior using the LDA (Latent Dirichlet Allocation) topic model, obtain the user-item probability matrix. Next, calculate user similarity based on the time when the user marked the item. On the basis of these, considering user relationship and calculate the user's final preference for the item and generate recommendations. Experimental results show that the performance is better than the traditional recommendation algorithm.