A hyperheuristic approach to examination timetabling problems: benchmarks and a new problem from practice

  • Authors:

  • Peter Demeester;Burak Bilgin;Patrick Causmaecker;Greet Berghe

  • Affiliations:

  • Information Technology, KAHO Sint-Lieven, Gent, Belgium 9000;Information Technology, KAHO Sint-Lieven, Gent, Belgium 9000;Computer Science and Information Technology, Katholieke Universiteit Leuven Campus Kortrijk, Kortrijk, Belgium 8500;Information Technology, KAHO Sint-Lieven, Gent, Belgium 9000

  • Venue:
  • Journal of Scheduling
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many researchers studying examination timetabling problems focus on either benchmark problems or problems from practice encountered in their institutions. Hyperheuristics are proposed as generic optimisation methods which explore the search space of heuristics rather than direct solutions. In the present study, the performance of tournament-based hyperheuristics for the exam timetabling problem are investigated. The target instances include both the Toronto and ITC 2007 benchmarks and the examination timetabling problem at KAHO Sint-Lieven (Ghent, Belgium). The Toronto and ITC 2007 benchmarks are post-enrolment-based examination timetabling problems, whereas the KAHO Sint-Lieven case is a curriculum-based examination timetabling problem. We drastically improve the previous (manually created) solution for the KAHO Sint-Lieven problem by generating a timetable that satisfies all the hard and soft constraints. We also make improvements on the best known results in the examination timetabling literature for seven out of thirteen instances for the To ronto benchmarks. The results are competitive with those of the finalists of the examination timetabling track of the International Timetabling Competition.