Teach and Explore: A Multiplex Information-guided Effective and Efficient Reinforcement Learning for Sequential Recommendation.

The article focuses on the limitations of current reinforcement learning-based sequential recommendation models, which fail to utilize supervision signals and auxiliary information, leading to slow convergence and limited exploration of user preferences. It mentions to overcome these challenges, the authors propose MELOD, a multiplex information-guided RL model, incorporating Teach and Explore components to accurately capture user preferences.