Narrowing the gaps in Concern-Driven Development

Concern-Driven Development (CDD) promises improved productivity, reusability, and maintainability because high-level concerns that are important to stakeholders are encapsulated regardless of how these concerns are distributed over the system structure. However, to truly capitalize on the benefits promised by CDD, concerns need to be encapsulated across software development phases, i.e., across different types of models at different levels of abstraction. Model-Driven Engineering plays an important role in this context as the automated transformation of concern-oriented models (a) allows a software engineer to use the most appropriate modeling notation for a particular task, (b) automates error-prone tasks, and (c) avoids duplication of modeling effort. The earlier transformations can be applied in a CDD process, the greater the potential cost savings. Hence, we report on our experiences in applying tool supported transformations from scenario-based requirements models to structural and behavioral design models during CDD. While automated model transformations certainly contribute to the three benefits mentioned above, they can also lead to more clearly and succinctly defined modeling activities at each modeling level and aid in the precise definition of the semantics of the used modeling notations.