In the context of Industry 4.0, companies must offer the market personalized systems (from goods to services). This customization is mainly supported by specific tools, called Configurators, which are based on a generic model formalizing the catalog and diversity proposed by a company, on a family of systems. The generic model or GM formalizes all the items (systems, subsystems, components, services, sub-modules, etc.) needed to build a system, in terms of nomenclature, compatibility (which items can be connected or associated to others) and options, which can be added to meet specific needs. Companies typically offer several variations of families of systems, all based on a common nomenclature. For example, a bicycle manufacturer will be able to offer its customers basic bicycles, road bikes, mountain bikes, children's bikes, women's bikes, etc. Two policies for formalizing knowledge in the form of GMs can thus emerge: 1) the GM formalizes all the knowledge and diversity proposed for all families of systems, 2) the GM formalizes only the knowledge and diversity specific to a family of systems; in this case, several GMs ranging from the most generic to the most specialized family cover the proposed diversity. It is thus possible to derive from a generic GM, the other more specific GMs. This specialization of generic models (from a basic bike to a competition road bike for children) can be achieved via a specialization mechanism between models. In this communication, we focus on this modeling problem by studying the complexity of the different models according to the chosen modeling policy. This complexity could be measured using the density of the model, its maximum degree, its number of vertices, etc.

Keywords
Configuration, Modeling, Specialization, Complexity assessment.