Abstract (EN)

Recent years have witnessed the increasing deployment of recommender systems as decision-support tools, application of recommender systems is ubiquitous in our daily life. As algorithms used in recommender systems become more inscrutable, some recommender systems remain blackbox for most people, which makes it difficult for people to fully trust such systems. Therefore, improving the user-recommender systems interactions is attracting increasing attention, one of the hot topics is explainable recommendation. In this paper, we investigate the possibility of providing context-aware explanations for recommendations and discuss the potential methods of evaluating such explanations.