Abstract:
UML design models are increasingly being used to support the design and evolution of large software
systems. Manual analysis of the entire UML model is impractical and is typically of little use. Since the
sizes of UML architectural models of systems themselves are increasing, efficient ways to understand,
analyze and maintain these models are becoming necessary and equally challenging. Development
of a coherent model to represent the design information scattered across various model elements can
be a promising solution. A slicing can then be performed on this model. Such model slicing can
assist in tedious and error-prone tasks such as model-based testing, understanding architectures, metrics
computation, early reliability prediction, and impact analysis. In this thesis, we first propose a coherent
metamodel for UML modelling artifacts and then propose slicing algorithms based on this.
