A study on the efficiency and accuracy of neural network model to optimize personalized recommendation of teaching content

How to fully explore the personality characteristics of student users and the implicit information in existing teaching resources, improve the existing teaching resources recommendation algorithm, and alleviate the problem of information overload and learning disorientation brought by massive teaching resources has become an important research topic in modern smart education and online education. In this paper, we first constructed a student cognitive diagnosis model based on graph convolutional neural network (GCN), and then constructed a teaching resource recommendation model based on convolutional joint probability matrix decomposition (CUPMF) based on cognitive diagnosis results, which successfully realized the optimization of the efficiency and accuracy of personalized recommendation of teaching content, and carried out an experimental evaluation of the model’s effect. Compared with other models, the cognitive diagnostic model in this paper achieves the best performance in ACC, AUC, and RMSE evaluation metrics, indicating that the combination of graph convolutional neural network model and traditional cognitive diagnostic model can be more effective in exploring the complex relationships among students, exercises, and knowledge points. Compared with other algorithms, the personalized recommendation algorithm for teaching resources designed in this paper has good performance in precision rate, recall rate and F1 index, and the mean value is lower in MAE index and CRR¯ standard deviation, which indicates that the model in this paper has the best recommendation effect and is closest to the ideal correct response rate. This paper provides a feasible path for optimizing the effectiveness of personalized recommendations using teaching content.