Tabu Search Techniques for Large High-School Timetabling Problems

No. 1: Agents, AI in Art and Entertainment, Knowledge Representation, and Learning

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 13

Stochastic Search

The high-school timetabling problem consists in assigning all the lectures of a high school to the time periods in such a way that no teacher (or class) is involved in more than one lecture at a time and other side constraints are satisfied. The problem is NP-complete and is usually tackled using heuristic methods. This paper describes a solution algorithm (and its implementation) based on Tabu Search. The algo- rithm interleaves different types of modes and makes use of an adaptive relaxation of the hard constraints. The implementation of the algorithm has been successfully experimented in some large high schools with various kinds of side constraints.