Personalized Recommendation Method for Training Courses on Enhancing Abilities of Staff of “Three Supports and One Assistance” with Big Data Analysis

With the rapid development of computer information technology such as machine learning, personalized recommendation systems have been widely used in the field of education and training. In view of the sparse interaction data between staff and training courses and the low matching degree of recommended courses, this paper adopted the combination model of Deep Q-Network (DQN) and collaborative filtering to study the personalized recommendation of training courses for “Three Supports and One Assistance”. First, this paper captured the interaction between staff and training courses through collaborative filtering model, and then used experiential playback to store and reuse historical interaction data for DQN model. It introduced a goal network to evaluate the goal Q-value to increase the efficiency of utilizing sparse data, and finally compared the performance of different personalized recommendation methods. The test outcome illustrated that the matching degree of DQN-Collaborative filtering model reached 96.5%, which was 4.1% higher than that of Collaborative filtering model. The response speed was only 30.5ms, the mean square error rate was only 0.02, and the accurate matching degree and good real-time performance were obtained. Research shows that this method has significant advantages in improving the accuracy and real-time performance of training course recommendations and has broad application prospects.