In the field of personalized learning path research for Chinese language education, traditional path planning lacks in-depth analysis of student behavior data, and it is difficult to provide real-time feedback and dynamic adjustment, resulting in low learning efficiency for students. This paper constructs an improved DQN model to collect various behavioral data of students during the learning process, provide necessary input for the DQN (Deep Q-Network) model, and convert these data into feature vectors as input for subsequent models; based on the collected learning data, a DQN model is constructed, and the state space, action space and reward mechanism are designed. During training, the model uses interaction with the environment to continuously optimize the Q-value function. The model adjusts the learning path based on the students' real-time learning performance. The model can continuously optimize the learning path to adapt to individual differences by continuously tracking the students' learning progress and feedback. Based on the students' learning behavior and mastery, the model assigns different learning tasks and content to other students. The experimental results show that the learning path planned by DQN in this paper can enable students to master 45 knowledge points in 8 hours of learning time, with a mastery accuracy rate of 0.90; when investigating students' satisfaction with the learning path planned by this model, 85%, 78% and 80% of them were satisfied or very satisfied with the learning progress, learning content and feedback mechanism respectively; the experimental results prove the effectiveness of this paper in the study of personalized learning path planning in Chinese language education.
Han et al. (Fri,) studied this question.