Perspectives of Agent-based Modeling and Simulation of Classroom Lessons




Konson, Grigoriy Rafaelevich
D.Sc. in Culture Studies, D.Sc. in Art Studies, PhD in Music History, Professor, the Advanced Education Department of the Landau Phystech School of Physics and Research, Director of the Humanities & Social Sciences Center, MIPT University, Moscow, Russia
konson.gr@mipt.ru

Kostin, Anton Aleksandrovich
PhD in Philosophy, Assistant Professor, the Humanities & Social Sciences Center, MIPT University, researcher, Sber Gamification Lab, Moscow, Russia
aa.kostin@phystech.edu


Abstract
In this talk we observe the state of the art in agent-based models and simulations of classroom lessons. The main purpose of classroom lessons modeling is to understand lesson interactions and have different validated scenarios of them. The most important argument to choose simulation instead of real experiment in this case is because of ethical reasons.
The empirical base usually is collected from informal discussions with individual teachers, focus group discussions with classes and lesson observations.
The most popular simulation method is agent-based modeling using NetLogo software which allows the use of various input and output factors (e.g. work difficulty, deadlines, completion etc.). For validation purposes the simulation usually is running more than 500 times and within the specified 95% confidence interval for all metrics.
We discuss two meaningful outcomes of these simulations: 1) the importance of complex agent architectures including emotional intelligence; 2) the importance of realistic environments including classroom space, class equipment and agents actions with them

Keywords: multi-agent interaction, simulation, elearning, digital education