Novel Local Searches for Finding Feasible Solutions in Educational Timetabling Problem

Any educational timetabling problem addresses the task of assigning a set of courses or exams to timeslots and rooms taking into account a set of constraints. Two types of constraints are usually imposed to the task (hard-compulsory and soft-non compulsory). In this paper, we address the feasibility problem, where the main concern is to find a timetable without any hard constraint violation. We propose some hill climbing-based local searches using what we call single and perturbation moves. The method is tested on a set of difficult instances dedicated for courses timetabling problem. The results are compared to other authors and it is shown to be capable to find the best solution found so far and improve the state-of-the art results for some instances.