|
Abstract:
|
This thesis proposes a set of model processing operations for manipulating architecture and design level software engineering models. The approach draws from well-established and widely used software modeling paradigms like class diagrams, statecharts, and interaction diagrams. The operations are based on the usually implicit dependencies that exist between models describing the same system from different perspectives, at different levels of abstraction or at different phases of evolution. The Unified Modeling Language (UML), as a widely adopted industrial standard providing a common design vocabulary, is chosen as the target modeling language.
The thesis outlines categories for model processing operations and defines them based on the UML metamodel. The thesis also studies how to combine the operations to form high-level model processing tasks. The presented operation categories are transformation operations, set operations, projection operations, and conformance operations. The main targets for applying the operations are assumed to be merging, slicing, synthesis, and checking of models. The presented approach aims towards supporting incremental model development, a faster and easier creation of models, improved model consistency and comprehension, and a better customization of model processing tools.
The thesis gives example usage scenarios for applying the operations and shows how they can be exploited in practice in the maintenance of real-life product platform architecture. Further, it shows that the operations can be implemented, integrated with a computer aided software engineering environment, and succesfully used during software engineering. The tools and techniques have been implemented and deployed in industrial settings. |