Paper Type
Complete
Paper Number
PACIS2026-1481
Description
To address the challenges of subjective and inefficient classroom behavior analysis in higher education, this study proposes the ATSBIM framework, integrating multimodal video analysis (YOLO, CNN) and educational theories. We automate the identification and quantification of teacher-student behaviors, and empirically test the impact of four core teacher behaviors on student engagement using 17,000 minutes of authentic classroom videos. Results confirm active teaching, proactive interaction, and instructional silence significantly boost engagement. This research bridges AI and IS education, providing a practical tool for classroom digital governance and theoretical support for educational digital transformation.
Recommended Citation
Yan, Mengling; Yang, Yiran; Zhang, Jiayuan; Zhang, Lifan; and Ye, Ziyi, "From Behavior Identification to Impact Validation: Exploring How Teacher Behaviors Shape Student Engagement through Multimodal Video Analysis" (2026). PACIS 2026 Proceedings. 5.
https://aisel.aisnet.org/pacis2026/is_education/is_education/5
From Behavior Identification to Impact Validation: Exploring How Teacher Behaviors Shape Student Engagement through Multimodal Video Analysis
To address the challenges of subjective and inefficient classroom behavior analysis in higher education, this study proposes the ATSBIM framework, integrating multimodal video analysis (YOLO, CNN) and educational theories. We automate the identification and quantification of teacher-student behaviors, and empirically test the impact of four core teacher behaviors on student engagement using 17,000 minutes of authentic classroom videos. Results confirm active teaching, proactive interaction, and instructional silence significantly boost engagement. This research bridges AI and IS education, providing a practical tool for classroom digital governance and theoretical support for educational digital transformation.
Comments
04-DigitalLearning