Formulation of model for decisional problems

In this paper the authors try to transfer some concepts of the theory of algorithms and computational complexity into the problem formulation technique. The first two sections are devoted to the formalization of what is a decision problem and what people-or better, the authors-think a decision problem formulation is. Concepts like decomposition and complexity are also discussed: the authors believe, in fact, that such concepts must be known by people who need to formulate a problem. The content of these first four sections are utilized in the following ones to build the ‘model network’, whose nodes are formulations of decision problems, with parametrized numerical data, and whose arcs (without loss of generality directed) represent the possibility of reformulating the starting model in terms of the final one. The concepts of algorithm, complexity and approximation are discussed in terms of the model network. Finally, the ‘skill’ and ‘learning capability’ of a computing system are introduced: such a computing system together with its skill and learning capability yields a ‘model handling system’ to support the decision making process.