Automating The Construction of Pedagogical Schedules

Abstract

Examination surveillance scheduling is an important administrative task in higher education institutions. As examination periods become more complex, manual scheduling requires significant effort and is often prone to errors. This work proposes an intelligent Exam Surveillance Scheduler that combines Integer Linear Programming (ILP) and Large Language Models (LLMs). The ILP component generates balanced and conflict-free invigilation schedules while satisfying constraints such as teacher availability, workload distribution, and examination coverage. The LLM component provides a natural language interface that enables users to query, modify, and analyze schedules efficiently. The system was developed and evaluated using real examination data from the Faculty of Mathematics and Computer Science at Mohamed Boudiaf University of M'Sila. The results demonstrate its effectiveness in reducing administrative workload and improving examination surveillance management through intelligent scheduling and user-friendly interaction.