- The paper introduces two algorithms that optimize team balance by integrating competencies, personality, and gender using a linear programming approach and an anytime heuristic.
- It demonstrates that the linear programming-based method excels in small instances while the heuristic achieves 75%-95% efficiency in larger, computationally challenging scenarios.
- Empirical evaluation on real-world data validates the models and supports their practical application via tools like EduTeams for effective educational team formation.
Introduction
In the field of cooperative learning, the formation of diverse and balanced teams is crucial for both academic accomplishments and individual knowledge acquisition. The strategies involving the assembly of efficient teams must account for the intricate interplay among competences, personality traits, and gender. This document explores the synergistic team composition problem (STCP), which seeks to create partitions wherein each team is not only balanced according to these factors but also capable of performing allocated tasks. To address the STCP, two distinct algorithms are introduced: a linear programming-based algorithm suitable for small instances and an anytime heuristic tailored for larger instances.
Related Work
Research in organizational psychology historically focuses on factors contributing to team success, lacking comprehensive computational models employed in classrooms. Although multiagent systems concentrate on agent competencies within teams, they tend to oversimplify skill representation. Furthermore, existing literature has not holistically considered all three factors simultaneously: competencies, personality, and gender when forming teams.
Synergistic Team Composition Model
The proposed model for STCP incorporates three dimensions of diversity: gender, personality, and competencies. In this context, the personality is measured using Post-Jungian methodology, while competencies are evaluated on a gradient, allowing for a fine-grained representation of individual skill levels. A task type within this framework defines the competencies required, their respective levels, and importance. The STCP aims to partition a classroom into teams such that each team maintains an even distribution of competencies, personalities, and gender, thus forming "synergistic" teams.
Algorithms to Solve STCP
Two algorithms are proposed for the STCP: STCPSolver utilizes integer linear programming (ILP) to generate an optimal solution by leveraging off-the-shelf ILP solvers. The SynTeam algorithm, designed as an anytime heuristic, offers quality solutions within constrained computational time frames which can be desirable for larger team sizes or when ILP-based approaches are computationally prohibitive. Comparative computational trials revealed that while STCPSolver is recommended for smaller team sizes due to its efficiency, SynTeam provides a noteworthy alternative for larger instances, striking a balance between solution quality and computational feasibility.
Empirical Evaluation and Application
An extensive empirical comparison of the proposed algorithms was conducted using real-world data. For smaller instances, STCPSolver's performance is commendable, but it incurs a prohibitively increasing runtime with larger team sizes and an increasing number of students. SynTeam, on the contrary, manages larger instances effectively, yielding approximate solutions with relative quality ranging from 75% to 95% of an optimal solution's value. The research also details a web application, EduTeams, which allows educators to apply the concepts of the STCP algorithmically for team composition in educational settings.
Conclusion and Future Directions
The STCP instigates fresh avenues for advanced and comprehensive models to accurately articulate varying determinants of team performance. Future endeavors may include dimensions like explicit expert preferences, consideration for multiple task types, and avenues for leveraging parallelism in algorithms to optimize performance. These paths reflect an evolution in forming teams that not only perform tasks with homogeneity but also enhance the collaborative learning experience.