Date: Monday, Jan 12 2026, 10:00am - 11:30am (PT)
Location: Stanford Graduate School of Education, Palto Alto, California

This presentation introduces the EDSI dataset, a large-scale multimodal, longitudinal classroom dataset to accelerate education R&D in the age of AI.  Collected from 300 4th-8th grade mathematics teachers, the dataset enables fine-grained analysis of classroom discourse through speaker identification, temporal tracking of interaction patterns, and linkage to survey and administrative data. The presentation addresses three fundamental questions about classroom dynamics: (1) Which students participate in classroom discourse, and how do patterns evolve over time? (2) How do observed interactions compare with student-reported experiences? (3) How do interaction features correlate with student achievement and teacher effectiveness? We conclude by outlining an agenda that leverages multimodal classroom data to advance research in education policy, teaching and learning, and their intersections with technology.

Jing Liu is the Founding Director of the Center for Educational Data Science and Innovation (EDSI) and an Associate Professor in Education Policy at the University of Maryland, College Park. While his background spans critical policy topics such as student absenteeism, school discipline, and teacher effectiveness, his recent work centers on the intersection of AI, data science, and education. Specifically, he focuses on building infrastructure for education R&D, developing and evaluating AI-powered edtech tools, and identifying the skills students need for a technology-driven future. In recognition of his contributions, Dr. Liu recently received the Early Career Award from the Association for Education Finance and Policy and the Exemplary Researcher Award from the University of Maryland.