A Heuristic Approach to Constraint Optimization in Timetabling

Timetabling is a difficult (NP-complete) problem and belongs to a general class of problems known as scheduling. Due to a variety of constraints typical in different timetabling environments, it has been difficult to develop a generic solution for timetabling. This paper is an attempt to define a generic computational model for examination timetabling for predefined constraints found in the problem, and proposes a heuristic method of developing an acceptable solution. The declarative nature of the developed constraints language (based on the structured query language) is utilized to construct constraints and specify the timetabling problem as a constraint satisfaction problem. A university examination timetabling problem is used to illustrate and test the model. 1 General Introduction This paper proposes a model for representing various constraints in a generic format that can be easily implemented in the differing timetable institutes. One reason why timetabling is a difficult problem is that there are many conflicting constraints that need to be resolved. Research done so far in this field has suggested various methods of building timetables that are context specific, as different timetabling institutes have many different types of constraints and data formats. This paper shows how the various constraints, in essence, are similar and that the seemingly different data have a generic format. A heuristic algorithm is applied to the model to generate a timetable. A prototype of the model was tested successfully with the University of the South Pacific's examination timetabling problem.