TGKT-Based Personalized Learning Path Recommendation with Reinforcement Learning

In recent years, emerging technologies represented by artificial intelligence have been widely applied in education, and online learning has transcended the limitations of time and space. How to recommend personalized learning paths for learners based on their characteristics has become a new issue. However, in the existing researches on personalized learning path recommendation, many researchers have not fully considered or only mined the initial personalized features of learners, and the process of learning path recommendations does not have good interpretability. In order to solve these issues, this paper proposes a model named TGKT-RL, in which we combine temporal convolutional network and graph attention network into the knowledge tracing model and use it as the environment of the reinforcement learning model. At the same time, we set learning goals for learners and adjust recommendation policy based on the states and rewards during the simulated learning process, ultimately recommending a personalized learning path. We conduct a series of experiments on two public knowledge tracing datasets, and the results show that our method achieves good performance and has good interpretability.