International Journal of Computer-Supported Collaborative Learning Education

Optimizing group formation with a mixed genetic algorithm: an empirical study in active reading using marker data

Changhao Liang, Yuko Toyokawa, Hiroaki Ogata
Springer Nature · 2025-08-01 · DOI 10.1007/s11412-025-09452-9

Abstract

Abstract Effective group formation is an indispensable yet challenging aspect of classroom-based collaborative learning. While existing group formation algorithms show promising computational performance in controlled settings, their practical impact on diverse, real-world classrooms remains underexplored. This paper presents a mixed genetic algorithm integrated into a data-driven learning platform designed to accommodate both homogeneous and heterogeneous student characteristics simultaneously. Implemented in a senior high school EFL classroom, the approach leverages active reading marker logs for data-driven grouping. It incorporates a WordCloud tool to enhance educators’ and learners’ understanding of group composition. Empirical results indicate that this system improves vocabulary learning, and the marker-based grouping strategies positively influence group learning dynamics. These findings underscore the algorithm’s practical relevance and highlight the benefits of interpretable, adaptive group formation methods for authentic educational contexts.

Read at publisher →