Explanation-Based Constraint Programming for a University Timetabling Problem

Abstract

Explanation-based constraint programming is a new way of solving constraint systems. It allows to propagate constraints of the problem, learning from failure and from the solver and finally allows to get rid of backtrack-based complete search by allowing more free moves in the search space. In this paper, we present our experience in using explanations within constraint programming: how to implement an explanation system, what to use explanation for solving a university timetabling problem. Beside classical uses, we are attempting to solve the problem with the class library of ILOG Solver [7] which leads to a new kind of explanation-based constraint programming.