Design of a multimodal data mining system for school teaching quality analysis

The multimodal teaching quality evaluation system realizes the intelligent analysis of teaching process through deep learning technology. The system collects classroom video, text and image data and constructs a deep neural network model for feature extraction and quality evaluation. The experiment collects data from 523 courses for testing, and the results show that the model accuracy rate reaches 90.3%, which is 23.6% higher than the traditional evaluation method. The system is deployed with distributed architecture, the average response time is 120ms, and the user satisfaction is 92.8%. The research results provide a new type of solution for teaching quality evaluation and have good application value.