Towards an Interoperable Metamodel Suite: Size Assessment as One Input

Abstract

In recent years, many metamodels have been introduced in the software engi- neering literature and standards. These metamodels vary in their focus across, for example, process, product, organizational and measurement aspects of software development and have typically been developed independently of each other with shared concepts being only ac- cidental. There is thus an increasing concern in the standards communities that possible con∞icts of structure and semantics between these various metamodels will hinder their widespread adoption. The complexity of these metamodels has also increased signiflcantly and is another barrier in their appreciation. This complexity is compounded when more than one metamodel is used in the lifecycle of a software project. Therefore there is a need to have interoperable metamodels. As a flrst step towards engendering interoperability and/or possible mergers between metamodels, we examine the size and complexity of various meta- models. To do this, we have used the Rossi and Brinkkemper metrics-based approach to evaluate the size and complexity of several standard metamodels including UML 2.3, BPMN 2.0, ODM, SMM and OSM. The size and complexity of these metamodels is also compared with the previous version of UML, BPMN and Activity diagrams. The comparatively large sizes of BPMN 2.0 and UML 2.3 suggest that future integration with these metamodels might be more di-cult than with the other metamodels under study (especially ODM, SSM and OSM).