Improving recommendation by deep latent factor-based explanation

The latent factor methods and explanation algorithms constitute the foundation of many advanced explainable recommender systems. However, interpreting the high-dimensional latent factors has not been sufficiently addressed and continuously becomes a challenging work. Besides, only a few works have researched the use of explanation to improve recommendations. In this paper, we propose a deep learning method that generates high-quality latent factor-based explanations and efficiently ameliorating recommendations. We conduct top- K items ranking experiment on two real-world datasets and show that our method outperforms nine currently state-of-theart recommender systems in five ranking metrics. Moreover, we conduct a qualitative and quantitative analysis of users’ latent factors and reveal that we continually offer the best latent representations.