Output details
11 - Computer Science and Informatics
University of York
Extensible Automated Constraint Modelling
<22>Reports the first-ever full implementation of a system
that automatically generates constraint models (i.e., sets of
constraints solvable by existing solvers) for problems. (Constraint
modelling was previously done only by humans, usually experts.)
Problems abstractly specified in the ESSENCE problem specification
language are refined to concrete models by the rule-based CONJURE
system described here. As the Introduction explicitly discusses,
the breakthrough was a new approach that enabled a complete
implementation of ESSENCE, whereas several prior approaches never came close.
The complete implementation facilitated a more extensive evaluation
that includes 32 problems and thousands of models. Conference acceptance rate: 24.8%.