A Collaborative Filtering Recommender Algorithm Based on the User Interest Model

To cope with the transfer of user interest and improve the accurate of prediction in recommender system, this paper proposes a Dynamic User-Interest-Model (DUIM). The model adopts the memory curve equation to address the influence of time factor. Users´ long term interest and short term interest can be embodied clearly in this model. Based on the model, the paper presents a novel collaborative filtering recommender algorithm (Model-based Collaborative Filtering Recommender Algorithm, MCF). Experiments prove that MCF gets better prediction and higher time efficiency compared with other similar algorithms.