Section: New Results

Learning Constraint Models

Participants : Nicolas Beldiceanu, Naina Razakarison, Helmut Simonis.

We designed a system which generates finite domain constraint models from positive example solutions, for highly structured problems. The system is based on the global constraint catalog , providing the library of constraints that can be used in modeling, and the constraint seeker tool , which finds a ranked list of matching constraints given one or more sample call patterns. We have tested the modeler with 230 examples, ranging from 4 to 6,500 variables, using between 1 and 7,000 samples. These examples come from a variety of domains, including puzzles, sports-scheduling, packing and placement, and design theory. Surprisingly, in many cases the system finds usable candidate lists even when working with a single, positive example.

The corresponding paper A Model Seeker: Extracting Global Constraint Models From Positive Examples [19] was published at the 18th International Conference on Principles and Practice of Constraint Programming (CP 2012 ).