Recommendations Based on Collaborative Filtering By Tag Weights
Feng Xiong, Yong Jian Liu, Qing Jie Xie · 2017
With the advent of social media and the exponential growth of information generated by online users, how to help users find knowledge from vast amounts of data has become the major problem to be solved. Probabilistic matrix factorization, which can handle massive amounts of data by learning low dimensional approximation matrices, but various works have ignored this relationships among users and resources. In this paper, a method is proposed to be based on tag weight of users and resources (TWPMF), which our use custom tags to more accurately identify the user interests and resource characteristics, so the influential neighbors are more accurately. First our use the operation weight and time factor to build user preference model to find the user neighbor set, and the resource neighbor set is obtained according to the custom label of the resource. Then those influential neighbors are successfully applied into the recommendation process based on probabilistic matrix factorization. The real-world data sets demonstrate that TWPMF algorithm can get more accurate rating predictions.