A structure-oriented approach to optimization modeling

The general approach to automating optimization modeling is to capture the combined knowledge of the domain expert and the optimization expert. The resulting knowledge base is capable of guiding decision makers through the process of optimization modeling to a solver. The major problem with this domain-specific approach is that decision problems may result from elements not contained in the knowledge base. This research explored a domain-independent, structure-oriented approach to assisting decision makers in formulating optimization decision models to the solver. The research analyzed the structural attributes of spreadsheet decision models in terms of their relevance to the optimization formulation. The resulting knowledge base, based on structural attributes, can help decision makers identify optimization-essential variables. This domain-independent knowledge base can complement a domain-specific knowledge base to increase the efficiency and generality of support for optimization modeling.