A recommendation system aims to provide personalized services or content by learning a huge amount of user behaviors with artificial intelligence. It enables a user-centric personalization in a heterogeneous setting, while the recommendation system can also be biased by focusing on only a limited viewpoint. A filter bubble, ie, a typical bias of a recommendation system, signifies the system utilizes filtered data according to the user; thus, it causes a side effect of confirmation bias and selective recognition, which in turn leads to a decrease in user satisfaction. In this study, we employed a systematic literature review method to specify the relationship between the bias of the recommendation system and user satisfaction. Based on a series of literature review protocols, we analyzed the articles from three angles:(i) the relationship between the bias of the recommendation system and user satisfaction and (ii) how to mitigate bias and improve user satisfaction. This research contributes to investigating a new angle, ie, the relationship between bias and user satisfaction, rather than a new technical algorithm of the recommendation system.