From theory to practice: solving a real-world multi-objective proctor assignment problem at a university in Istanbul


GEÇİCİ BİRKAN E., IŞIK E. E., Simsir Gunes M., GÜNEŞ M.

INTERNATIONAL JOURNAL OF INTELLIGENT COMPUTING AND CYBERNETICS, 2025 (ESCI, Scopus) identifier

Özet

PurposeThis study aims to address the exam proctor assignment problem faced by university departments by developing a fair, efficient, and practically applicable scheduling model. The issue involves multiple objectives, including balanced workload distribution, minimizing back-to-back assignments, reducing walking distances between rooms, and respecting individual preferences.Design/methodology/approachA multi-objective mathematical model is proposed and solved using an Interactive Fuzzy Goal Programming (IFGP) approach, which enables incorporating decision-makers' preferences and handling soft constraint violations. Additionally, classical fuzzy goal programming and a penalty-cost-based mixed-integer linear programming (MILP) reformulation were employed to evaluate alternative solution strategies. Real-world data from a university in Istanbul were used to test the model under different staffing scenarios.FindingsThe proposed method generated equitable and operationally feasible schedules in both low- and high-staffing scenarios. The model effectively minimized consecutive assignments and travel distances while also accommodating individual proctor requests. The results demonstrate that IFGP offers greater flexibility and interpretability than the other tested approaches.Originality/valueUnlike previous studies that focus on centralized or large-scale invigilation systems, this research provides a department-level, real-life implementation of a multi-objective proctor assignment model. The integration of soft constraints, real decision-maker interaction, and comparative evaluation of alternative solution methods enhances both the academic and practical value of the study. The model can be adapted for use in various institutional contexts and extended for dynamic scheduling environments.