Emotion Classification and Achievement of Students in Distance Learning Based on the Knowledge State Model

Abstract

Since the outbreak of COVID-19, remote teaching methods have been widely adopted by schools. However, distance education can frequently lead to low student emotional engagement, which can not only cause learning burnout, but also weaken students’ interest in online learning. In view of the above problems, this study first proposed a learner knowledge state model that integrates learning emotions under the background of digital teaching to accurately describe the current learning state of students. Then, on the basis of the public face dataset lapa, we built an online multi-dimensional emotion classification model for students based on ResNet 18 neural network. Experiments showed that the method has an average recognition accuracy of 88.76% for the four cognitive emotions of joy, concentration, confusion, and boredom, among which the accuracy of joy and boredom is the highest, reaching 96.3% and 97.0% respectively. Finally, we analyzed the correlation between students’ emotional classification and grades in distance learning, and verified the effectiveness of the student’s emotional classification model in distance learning applications. In the context of digital teaching, this study provides technical support for distance learning emotion classification and learning early warning, and is of great significance to help teachers understand students’ emotional states in distance learning and promote students’ deep participation in the distance learning process.