University Course Timetabling with Soft Constraints

An extension of constraint logic programming that allows for weighted partial satisfaction of soft constraints is described and applied to the development of an automated timetabling system for Purdue University. The soft-constraint solver implemented in the proposed solution approach allows constraint propagation for hard constraints together with preference propagation for soft constraints. A new repair search algorithm is proposed to improve upon initially generated (partial) assignments of the problem variables. The model and search methods applied to the solution of the large lecture room component are presented and discussed along with the computational results.