Computer Science · Information Systems
Matrix Factorization Based Models Considering Item Categories and User Neighbors
See on the mapAbstract
Matrix factorization is a popular collaborative filtering method for recommendation techniques with predictive accuracy and good scalability. In this paper, we propose two models on the basis of basic matrix factorization, namely CW-MF, NICW-MF. CW-MF considers user's preference on item categories and NICW-MF takes into account the impact of user's neighbors to minimize the preference between user and his neighbors. We conduct empirical experiments on MovieLens dataset, and results show that our two models perform well.