Collaborative filtering recommendation algorithm based on shift of users´ preferences

The weight of all users´ score is the same in traditional collaborative filtering recommendation algorithm, and it doesn´t consider the shift of users´ preferences with time, so recommendation quality is poor. In order to avoid the problems above, a novel collaborative filtering algorithm based on shift of users´ preferences is presented: The method adjusts the weight of users´ score according to time, improves users´ similarity with a gradual forgetting function, and considers the impacts on similarity between users brought by the shift of users´ preferences, then clusters users in terms of their features, reduces chosen space of the nearest neighbors. The experiment result shows that this method has better recall and precision than traditional collaborative filtering recommendation algorithm, and it can effectively improve recommendation quality.