Does It Look Sequential? An Analysis of Datasets for Evaluation of Sequential Recommendations

Sequential recommender systems are an important and demanded area of research. Such systems aim to use the order of interactions in a user’s history to predict future interactions. The premise is that the order of interactions and sequential patterns play an essential role. Therefore, it is crucial to use datasets that exhibit a sequential structure to evaluate sequential recommenders properly.