Introduction to the Special Issue on the Integration of AI and OR Techniques in CP for Combinatorial Optimization (CPAIOR 2005)

Abstract

The Second International Conference on Integration of Artificial Intelligence (AI) and Operations Research (OR) Techniques in Constraint Programming (CP) for Combinatorial Optimization Problems (CPAIOR 2005) was held in Prague, Czech Republic, during May 30–June 1, 2005. The interest of the research community in this conference is witnessed by the high number of submissions received, almost reaching 100 papers submitted to CPAIOR 2005. From these submissions, we chose 26 papers to be published in full in the proceedings and only five papers to be extended into a journal form, and submitted to this Special Issue. These papers are explained here in greater depth than was possible in the conference proceedings. The first three papers have a theme in common: global constraints and filtering algorithms. Global constraints play a significant role in Constraint Programming and in hybrid approaches. They are a vehicle for embedding OR techniques in CP and they are crucial to the solution of hard combinatorial problems. The first paper in this group is entitled Filtering Algorithms for the NValue Constraint by Bessiere et al. The paper shows some approximations for computing a bound on the total number of different values an array of variables can assume. The approximations are compared as far as filtering and complexity are concerned. The second paper in the global constraint group is entitled A Cost-Regular Based Hybrid Column Generation Approach by Demassey et al. It proposes a costbased version of the regular constraint, where the longest and the shortest path are computed in the layered graph of the regular constraint. The constraint is particularly useful for solving sub-problems in hybrid Column Generation approaches. The constraint is applied to Employee Timetabling Problems to show its effectiveness.