Introduction to the special issue on abstraction and automation in constraint modelling

Constraints are powerful and natural means of knowledge representation and inference, and can solve a wide range of combinatorial problems. Constraint solving of a combinatorial problem proceeds in two phases, in first phase, the problem is modeled by a set of constraints on decision variables that its solutions must satisfy and in second phase, a constraint solver is used to search for solutions to the model. The way to improve usability is by extending constraint technology to enable models to be formulated at a higher level of abstraction. Automation can also aid the modeling process by transforming a constraint model into one that can be solved more effectively. Such transformations include adding implied constraints, recognizing symmetries in models, adding constraints to exploit dominance in optimization problems, removing propagation-redundant constraints, and creating relaxed versions of the initial problem.