Tutorial: Sequence-aware recommender systems

Abstract

Recommender systems are widely used in online applications to help users find items of interest and help them deal with information overload. In this tutorial, we discuss the class of sequence-aware recommender systems. Differently from the traditional problem formulation based on a user-item rating matrix, the input to such systems is a sequence of logged user interactions. Likewise, sequence-aware recommender systems implement alternative computational tasks, such as predicting the next items a user will be interested in an ongoing session or creating entire sequences of items to present to the user. We propose a problem formulation, sketch a number of computational tasks, review existing algorithmic approaches, and finally discuss evaluation aspects of sequence-aware recommender systems.