Explainable matrix factorization with constraints on neighborhood in the latent space

Abstract

Nowadays, recommender systems are widely used to solve the problem of information overload in modern society. And most of the previous studies focus overwhelmingly on high accuracy in the recommender systems. But in a real system, the high accuracy does not always satisfy overall user experience. The explainability has a great impact on the user experience. We mainly focus on the explainability of recommender systems in this paper. To the best of our knowledge, it is the first time that the neighborhood information in the latent space is integrated into the Explainable Matrix Factorization. We change the method of calculation of the explainability matrix and consider the neighbors’ weight to further improve performance. We use the benchmark data set (MovieLens) to demonstrate the effectiveness of the proposed Neighborhood-based Explainable Matrix Factorization. And the result shows a great improvement for accuracy and explainability.