A Hierarchical Integer Linear Programming Approach for Optimizing Team Formation in Education

TL;DR

This paper introduces a flexible, modular integer linear programming model for optimizing team formation in education, addressing limitations of previous methods by incorporating student agency and varying educational needs.

Contribution

It presents the EDU-TF model, a novel, adaptable optimization framework tailored to educational contexts, allowing for customizable objectives and improved student preference accommodation.

Findings

Our approach outperforms heuristic teacher-assigned teams in real-world datasets.

The modular model adapts to various educational settings and objectives.

Results demonstrate better alignment with student preferences.

Abstract

Teamwork is integral to higher education, fostering students' interpersonal skills, improving learning outcomes, and preparing them for professional collaboration later in their careers. While team formation has traditionally been managed by humans, either instructors or students, algorithmic approaches have recently emerged to optimize this process. However, existing algorithmic team formation methods often focus on expert teams, overlook agency in choosing one's teammates, and are limited to a single team formation setting. These limitations make them less suitable for education, where no student can be left out, student agency is crucial for motivation, and team formation needs vary across courses and programs. In this paper, we introduce the EDUCATIONAL TEAM FORMATION problem (EDU-TF), a partitioning optimization problem model tailored to the unique needs of education, integrating both teacher and student requirements. To solve EDU-TF, we propose a modular optimization approach, one of the first to allow the flexible adjustment of objectives according to educational needs, enhancing the method's applicability across various classroom settings rather than just research environments. Results from evaluating ten strategies derived from our model on real-world university datasets indicate that our approach outperforms heuristic teacher-assigned teams by better accommodating student preferences. Our study contributes a new modular approach to partition-based algorithmic team formation and provides valuable insights for future research on team formation in educational settings.