A Constraint Seeker: Finding and Ranking Global Constraints from Examples
Résumé
In this paper we describe a Constraint Seeker application which provides a web interface to search for global constraints in the global constraint catalog, given positive and negative, fully instantiated (ground) examples. Based on the given instances the tool returns a ranked list of matching constraints, the rank indicating whether the constraint is likely to be the intended constraint of the user. We give some examples of use cases and generated output, describe the different elements of the search and ranking process, discuss the role of constraint programming in the different tools used, and provide evaluation results over the complete global constraint catalog. The Constraint Seeker is an example for the use of generic meta-data provided in the catalog to solve a specific problem.