Co-evolution of (information) system models

Abstract

Information systems' modelling is based on separation of concern such as separation into facets or viewpoints on the application domain from one side and separation of aspects (structuring, functionality, interactivity, distribution, architectural components) from the other side. Facets and aspects are typically specified through different models that must be harmonised and made coherent. Such varieties of models are difficult to handle, to evolve, to maintain and to use. Most design methodologies adopt the master-slave principle in order to handle the coherence of such model assemblies by assigning one model to be the master and mapping the master to slave models. Moreover, these models diagrams are typically not developed from scratch. They are incrementally completed step by step depending on the modelling methodology. Models evolve during development and are not independent, are interrelated, and in most applications also intertwined. Their interrelationships are often not made explicit and impose changes resulting in inconsistencies to other models due to the variety of models. Therefore, this paper introduces the theory of model suites as a set of models with explicit associations among the models. Model suites are based on explicit controllers for maintenance of coherence, apply application schemata for their explicit maintenance and evolution, use tracers for establishment of their coherence and thus support co-evolution of information system models. The excitability is captured by integrating model suites and MetaCASE formalisms, exploring the (modelling) method engineering and tool generation required for multi-model development. © 2010 Springer-Verlag Berlin Heidelberg.