The Diagnosis Model of Students’ Cognitive Level Based on Deep Learning

In order to guide students’ personalized learning based on personalized characteristics, the author proposes a deep learning-based diagnostic model for students’ cognitive level. First, initialize a knowledge point mastery vector for students, and each dimension represents the mastery degree of the corresponding knowledge point. Then use the deep learning model to learn the parameters of the project reflection theory from the mastery vector, topic text, and topic inspection knowledge points: The potential characteristics of students, the difficulty of the questions, and the degree of distinction between the questions. Finally, the item response function in item reflection theory is used to predict students’ scores. Experimental results show that: Extended experiments are carried out on the performance prediction task under different data sparsity, and the dataset is divided into training and testing datasets with different ratios: 60, 70, 80, 90%, compared with the baselines, especially IRT, MIRT, and DIRTNA, DIRT performs the best, indicating that the attention mechanism-based LSTM is effective for mining the question text, and DIRT can provide more accurate diagnostic results to augment traditional IRT. Conclusion: Extensive experiments on real datasets clearly demonstrate the effectiveness and interpretability of DIRT.