A general approach to constraint solving for declarative modeling domain

Abstract

Research in declarative modeling started some years ago, and important progress has been accomplished and different orientations have been studied. However, some fundamental problems are not exhaustively explored: the role of constraint solvers in maintaining the scene; detecting the incoherence and contradictions between constraints; reducing the number of generated solutions by dynamically adding new constraints. Our main contribution concerns declarative modeling with constraints. We have developed a constraint solver called ORANOS that offers an extended model of constraint satisfaction problems. The solver supports two independent domains of artificial intelligence research: hierarchical constraints and dynamical constraints. The former offers efficient solving techniques for over-constrained problems; the latter allows development of interactive applications. These essential features allow the solver to extend the range of declarative modeling applications.