This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
MathBuddy: A Multimodal System for Affective Math Tutoring
Created by
Haebom
Author
Debanjana Kar, Leopold B oss, Dacia Braca, Sebastian Maximilian Dennerlein, Nina Christine Hubig, Philipp Wintersberger, Yufang Hou
Outline
This paper points out the limitations of existing LLM-based interactive learning systems, which fail to consider students' emotional states. We present MathBuddy, an emotion-aware mathematics learning system that models students' emotions and dynamically adjusts teaching strategies. MathBuddy identifies students' emotions through conversational text and facial expressions, and synthesizes these data to induce emotionally appropriate responses from the LLM tutor. Using automated evaluation metrics and user research to evaluate MathBuddy's effectiveness, we confirmed significant performance improvements over existing systems.
Takeaways, Limitations
•
Takeaways:
◦
It shows that considering students' emotions in LLM-based education systems is important for improving learning effectiveness.
◦
Demonstrating the effectiveness of a multimodal emotion recognition approach combining conversational text and facial expression analysis.
◦
Presenting a new direction for developing emotion recognition-based training systems and verifying practical performance improvements.
•
Limitations:
◦
The scope of the study was limited to mathematics learning. Further research is needed to determine generalizability to other subjects and learning fields.
◦
Further validation is needed on the accuracy of facial expression analysis and its generalizability to various emotional expressions.
◦
Consideration should be given to the generalizability of the results based on the scale of the user study and participant characteristics.
