A meta constraint logic programming architecture for temporal reasoning



AUTHOR: Michela Milano

AFFILIATION: DEIS University of Bologna

ABSTRACT:

Constraint Logic Programming (CLP) is a powerful programming paradigm combining the advantages of Logic Programming and the efficiency of constraint solving. I will focus on Constraint Logic Programming on finite domains, CLP(FD). CLP(FD) is a suitable paradigm for dealing with constraint-based temporal frameworks. However, commercial constraint solvers present some limitations. First, they have an embedded arc consistency propagation algorithm which cannot be changed by the user in accordance with the application to be solved. Second, CLP solvers are not able to deal with qualitative temporal constraints. I will show how to overcome these limitations by using a CLP meta architecture where higher levels reason on an explicit or implicit representation of the underlying level constraints. I will present two applications of the architecture. The first extends the expressive power of CLP and deals with qualitative and quantitative temporal reasoning. The second allows to incrementally achieve higher levels of consistency by combining weaker consistency algorithms. The architecture presented inherits the modularity and flexibility of meta architectures and the efficiency of constraint solving.