Local Search and Constraint Programming for a Real-World Examination Timetabling Problem

Abstract

We investigate the examination timetabling problem in the context of Italian universities. The outcome is the definition of a general problem that can be applied to a large set of universities, but is quite different in many aspects from the classical versions proposed in the literature. We propose both a metaheuristic approach based on Simulated Annealing and a Constraint Programming model in MiniZinc. We compare the results of the metaheuristic approach (properly tuned) with the available MiniZinc back-ends on a large set of diverse real-world instances.