Dynamic Monitoring Model of English Classroom Attention Driven by Generative Artificial Intelligence

Against the backdrop of global educational informatization, the dynamic monitoring of student attention in English classrooms is a critical factor in improving teaching quality. However, traditional methods primarily rely on teachers' subjective observations and static evaluations, which are insufficient for capturing the dynamic fluctuations of attention in multi-task scenarios such as listening comprehension and oral dialogue. Furthermore, these conventional approaches are limited by their single monitoring dimension and disconnection from teaching practice. To address these limitations, this study aims to construct a dynamic attention monitoring model for English classrooms, driven by generative artificial intelligence (GenAI). Utilizing the student engagement dataset (SED), the study integrates multimodal data, including facial micro-expressions, voice interactions, and learning behaviors. Feature optimization is achieved through Z-score standardization and cosine similarity weighting.