The versatility of using explanations within constraint programming

Abstract

Constraint programming is a research topic benefiting from many other areas: discrete mathematics, numerical analysis, artificial intelligence, operations research, and formal calculus. It has proven its interest and its efficiency in various domains: combinatorial optimization, scheduling, finance, simulation and synthesis, diagnosis, molecular biology, or geometrical problems. However, some limitations and difficulties remain: designing stable and generic algorithms, handling dynamic problems, opening constraint programming to non-specialists, etc.<br /><br />In this document, we advocate the use of explanations within constraint programming. Our aim is two-fold: drawing the big picture about explanations (definition, generation, management and use) and showing that they can help address several issues in constraint programming. We also introduce a new general explanation-based search technique that has been successfully used to design new efficient algorithms. Finally, current open issues and research topics in this field are presented.