智慧教学背景下的教学质量预测——基于多模态与复杂网络的应用  

Prediction of teaching quality in context of smart teaching:research and application based on multi-modal and complex networks

作　　者：钟元权 ZHONG Yuanquan(Office of Teaching Affairs,Anhui Wenda Information Engineering College,Hefei 231201,China)

机构地区：[1]安徽文达信息工程学院,安徽合肥231201

出　　处：《南昌工程学院学报》2024年第6期82-90,共9页Journal of Nanchang Institute of Technology

基　　金：安徽省教育厅重点教育教学管理项目(2022jxgl027)。

摘　　要：现有的教学质量预测方法主要依赖学生成绩等单一模态数据,未能充分挖掘课堂复杂网络结构,导致对多维互动关系理解不足,影响预测精度。通过引入复杂网络理论,构建教学互动网络,以揭示课堂互动规律及其关键影响因素,提升教学质量预测的精度和鲁棒性;利用成绩管理系统和智能摄像头收集多模态学习数据,采用小波变换对原始信号进行去噪,并引入注意力机制筛选输入特征,分析学生、教师和学习资源之间的关系;通过融合多模态数据,为各模态分配不同权重,整合教育信息并突出关键信息,构建了基于多模态与复杂网络拓扑结构的教学质量预测模型。该模型在测试中的平均准确率达到94.16%,F值为90.60%,AUC值为0.975,均方误差为0.271,显著提升了教学质量预测的准确性与可靠性。The existing teaching quality prediction methods primarily rely on single-modal data such as student grades,failing to make full use of the complex network structure of classrooms,which leads to an insufficient understanding of multi-dimensional interactive relationships and affects prediction accuracy.By introducing complex network theory,a teaching interaction network is constructed to reveal classroom interaction patterns and key factors,enhancing the precision and robustness of predictions.Multi-modal learning data are collected using grade management systems and smart cameras.Wavelet transform is applied to denoise the raw signals,and attention mechanisms are introduced to filter input features.The relationships between students,teachers,and learning resources are analyzed to construct the topology of a complex network.This method integrates multi-modal data by assigning different weights to each modality,consolidating educational information,and highlighting key information.Test results show that the method achieved an average accuracy of 94.16%,an F1 score of 90.60%,an AUC value of 0.975,and a mean squared error of 0.271 in the first test,which significantly improves the accuracy and reliability of the prediction of teaching quality.

关 键 词：教学质量预测 复杂网络 注意力机制 多模态数据 深度神经网络 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]