Assigning Satisfaction Values to Constraints: An Algorithm to Solve Dynamic Meta-Constraints

Abstract

The model of Dynamic Meta-Constraints has special activity constraints which can activate other constraints. It also has meta-constraints which range over other constraints. An algorithm is presented in which constraints can be assigned one of five different satisfaction values, which leads to the assignment of domain values to the variables in the CSP. An outline of the model and the algorithm is presented, followed by some initial results for two problems: a simple classic CSP and the Car Configuration Problem. The algorithm is shown to perform few backtracks per solution, but to have overheads in the form of historical records required for the implementation of state.