Reinforcement Learning for Online Learning Recommendation System

In the learning environment, individual learner requires flexible and suitable learning processes. Online learning should be able to recommend appropriate learning objects (LOs) to an individual in real-time. Most of the existing approaches of online learning recommendation systems are based on collaborative filtering methods. Such methods have a limitation on realtime adaption and require the prior knowledge of students and LOs. Therefore, this study proposes a real-time recommendation method which is suitable for flexible and complex environments. The proposed method is based on Reinforcement Learning problem. The method is able to explore the environment to get information and exploit the information to make a decision. We evaluate the proposed method with the real world data. We vary e-greedy, the learning rate, and the discount rate for a tradeoff between the exploration and exploitation.