The Application of a Multimodal Analysis System Based on the COZE AI Agent in Education

Personalized learning has garnered significant attention, yet research on the impact of teaching methods on learners' emotional and cognitive states is limited. This study integrates multimodal learning analytics (MLA) with AI agent technology to examine how Confucian and Socratic teaching methods affect learners' emotions and cognition. Using low-cost, portable devices, we collected multimodal data, including eye-tracking and facial expression data, from undergraduate students exposed to both teaching methods. We developed AI agents on the COZE platform to analyze these data, assess learners' emotional and cognitive states, and deliver personalized feedback. Results show that the Confucian method promotes emotional stability and sustained attention, while the Socratic method encourages deeper cognitive engagement but may lead to emotional fluctuations. This research underscores the potential of MLA and AI agents in optimizing teaching methods, contributing to personalized learning and adaptive educational technologies.