Prediction of teaching quality in the context of smart education: application of multimodal data fusion and complex network topology structure
Abstract The existing teaching quality prediction methods only rely on single modal data such as students’ scores, and do not fully mine the complex network structure of the classroom, resulting in insufficient understanding of multidimensional interaction relationships and affecting the prediction...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-03-01
|
| Series: | Discover Artificial Intelligence |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44163-025-00240-w |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The existing teaching quality prediction methods only rely on single modal data such as students’ scores, and do not fully mine the complex network structure of the classroom, resulting in insufficient understanding of multidimensional interaction relationships and affecting the prediction accuracy. This article constructed a teaching interactive network by applying complex network theory, and used complex network analysis to reveal classroom interaction rules and key factors, improving the accuracy and robustness of teaching quality prediction. Firstly, the article utilized a score management system and intelligent cameras to collect multimodal learning data; then, the original signal was decomposed into detail layers and approximation layers of different frequencies using wavelet transform, and high-frequency detail layers were processed for denoising through threshold functions; afterwards, in order to reduce the network size and parameters, attention mechanism was applied to screen the input features of the neural network; using students, teachers, and learning resources as network nodes, the connections between nodes were analyzed and a complex network topology structure was constructed; then, attention mechanism was used to fuse multimodal data, assigning different weights to each modality, integrating educational information, and highlighting key information in the prediction task. Through testing the collected multimodal data such as grades, classroom behavior, and psychological characteristics, it was found that this article’s method could optimize teaching strategies and improve teaching effectiveness. The prediction model based on attention mechanism optimized deep neural network achieved an average accuracy of 94.16% in the first test; the average F1 score was 90.60%; the AUC (Area Under the Curve) value was 0.975; the average mean square error was 0.271. The attention mechanism optimized deep neural network teaching quality prediction model studied in this article has improved the accuracy and reliability of prediction, achieving comprehensive and accurate prediction of teaching quality. |
|---|---|
| ISSN: | 2731-0809 |