The Sequential Recommender is an important type of recommendation system. Its main task is to predict the user's next behavior based on the user's historical behavior sequence. It attempts to understand and model the temporal information of user behavior, the interaction between users and items, and the changes of these factors over time, so as to provide users with more accurate recommendations that meet their current and recent preferences. For example, a user first browsed computers and keyboards on an e-commerce platform, and then browsed mice. The Sequential Recommender will predict that the user may be interested in computer-related accessories such as computer speakers based on this series of behaviors, and recommend such products to the user.

In early studies, some methods based on simple models such as Markov chains were used for sequence recommendation. Later, with the rise of deep learning technology, sequence recommendation methods based on deep learning models such as recurrent neural networks (RNN), long short-term memory networks (LSTM), gated recurrent units (GRU), and Transformers continued to emerge.

The paper published in IJCAL 2019Sequential Recommender Systems: Challenges, Progress, and Prospects"This paper provides a comprehensive overview of sequential recommendation systems, and systematically summarizes and analyzes their characteristics, challenges, research progress, and important research directions.