Collaborative Filtering Recommendation with Fluctuations of User’ Preference
Jing Yu, Jinaiing Shi, Yunwen Chen, Daqi Ji, Wenhai Liu, Zhijun Xie, Kai Liu, Xue Feng
Abstract
Traditional recommendation systems (RSs) recommend users to personal items according to their pre-behaviors to infer users' personal preference. Although researchers have made some improvements, which take time factor into traditional recommendation systems, on catching changes of users' interest, it is difficult to always compute the users' preference fluctuations. Therefore, to solve the mentioned above problems, we add personal preference fluctuations into traditional collaborative filtering systems. In this paper, we will quantify user's preference change. Besides, it also adapts recommending in the opposite direction, that is to say, we will recommend items which dissimilar users have rated to the pointed user. The experiment of fluctuation method run on MovieLens 1M and MovieLens latest small, show that the method including changes of personal preference is better than other methods.