ON METHODOLOGY FOR THE ELABORATION OF A THEORETICAL MODEL
Pierre Delattre · International Journal of General Systems · 1975
With a set of experimental data, there are various methods of searching for an interpretative theoretical model. With the most frequently used, stress is put on the more or less precise correspondence between computed and measured variables rather than on epistemological consistency. The models thus obtained often rely upon imprecise or even quite implicit concepts, and they can be called ”closed upon themselves,“ as there is no possibility of structural extrapolation towards new models. There is only a possibility of an “opening” when the search for models is made in a framework of well axiomatized theory which plays the role of a metal-model with respect to particular cases. Some examples belonging to linear transformation systems serve as an illustration of the appearance of such an “opening.” Systems of this type are based on state description2,3. Elements are distributed indifferent equivalence classes according to their properties, i.e. their state, The search for an interpretative model thus consists of finding, from the observed contents of classes with respect to time, the transition probabilities from state to state.