Inferring constraint types in constraint programming

Abstract

Capturing constraint structure is critical in Constraint Programming to support the configuration and adaptation of domain filtering algorithms. To this end, we propose a software model couplinga relational constraint language, a constraint type inference system, and an algorithm configuration system. The relational language allows for expressing constraints from primitive constraints; the type system infers the type of constraint expressions out of primitive constraint types; and the configuration system synthesises algorithms out of primitive routines using constraint types. In this paper, we focus on the issue of constraint type inferencing, and present a method to implement sound and extendible inference systems.