Student behavior analysis to measure engagement levels in online learning environments

Area of Science:

• Computer Science
• Educational Technology
• Human-Computer Interaction

Background:

• The COVID-19 pandemic highlighted the importance of smart online learning systems.
• Measuring student engagement is critical for adaptive learning environments.
• Existing methods for measuring engagement are often limited or expensive.

Purpose of the Study:

• To propose and validate a novel automatic multimodal approach for real-time student engagement measurement.
• To address the challenge of accurately assessing student engagement based on behavior and emotions.
• To develop a cost-effective solution using readily available equipment.

Main Methods:

• A multimodal approach combining facial expression analysis, keyboard keystroke logging, and mouse movement tracking.
• Real-time data analysis of student behavior.
• Validation through experiments using newly created student engagement datasets.

Main Results:

• The multimodal approach achieved the highest accuracy of 95.23%.
• The mean square error (MSE) was recorded as low as 0.04.
• Experimental validation demonstrated the effectiveness of combining multiple behavioral modalities.

Conclusions:

• The proposed automatic multimodal approach provides robust and accurate real-time measurement of student engagement.
• This method offers a cost-effective solution for enhancing smart online learning systems.
• Future research can build upon this multimodal strategy for more sophisticated educational technology.