When Attention Turn To Be Explanation. A Case Study in Recommender Systems

Recent studies in deep learning aim to understand how intermediate representations learned from attention mechanisms motivates the decisions of a predictive model and, consequently, provides information on the model's decision-making process. In fact, while the effectiveness of attention is a well-established topic, the power of attention coefficients to express explanations remains a somewhat controversial issue in the literature. In this work, we empirically evaluate the possibility of using attention coefficients to obtain faithful explanations for recommender systems. In particular, after showing how to use attention for explaining recommendations, we examine the robustness of our proposal.