New checker for constraint network solutions

Khalid Haddouch, Karim El Moutaouakil

Abstract: A large variety of real-world problems in artificial intelligence can be viewed as Constraint Satisfaction Problems CSPs, Max-CSP, Weighted CSP, and other extensions. All resolution methods generate a solution. But, in some cases, these solutions are not correct and it is necessary to specify the detail of the solution(assigned values for variables and violated constraints). In this paper, we propose a new checker that can be used, as reference, to verify any solution generated by any solvers of CSP, Max-CSP and WCSP. Noted that, the one of the most advantages of this checker that can specify, exactly, the violated constraints as well as variables and values that causes this violation. In this regard, A set of theorems, propositions, examples and experimental results are integrated and discussed in order to validate the performance of the proposed checker.