UNIFICATION OF LINEAR-PROGRAMMING WITH A RULE-BASED SYSTEM BY THE POST-MODEL ANALYSIS APPROACH

Abstract

We attempt to unify a linear programming model with a rule-based system via overlapped decision variables. Since a rule base can be regarded as an acyclic AND/OR digraph, the unified model is virtually a multiobjective decision making problem that considers both numeric and symbolic objectives and decision variables. To solve this problem, the Post-model Analysis approach is adopted. An essential feature of the Post-model Analysis is that it supports nondominated tradeoffs among the numeric and symbolic objectives. So, this paper proposes a set of methods for computing the nondominated solutions. A prototype UNIK-PMA that realizes the idea is developed as a subsystem of the UNIK (UNIfied Knowledge) system.