Teaching Content Recommendation Strategy based on Neural Network Unit Rating

With the continuous improvement of the online teaching platform, data are accumulated to provide the main basis for tailor-made teaching based on students' personality characteristics. Process assessment data and learning evaluation data are collected, a unit evaluation strategy based on the long and short-term memory neural network LSTM is given, and a recommendation strategy for students' pre-review content is given based on the evaluation strategy. The strategy is applied to the knowledge recommendation problem of the basic artificial intelligence experimental course, and the results of the strategy are analyzed and compared with the evaluation strategy results of the traditional Convolutional Neural Network (CNN) and Support Vector Machine (SVM). The data show that the recommendation accuracy of the LSTM-based multi-attribute evaluation strategy is relatively high in recommending knowledge points of neighboring chapters, and the error is low when compared with the actual student questionnaire data. In recommending knowledge points for chapters that span a wide range of chapters, there is not much difference between the two algorithms.