Model Refinement in the Model Driven Architecture Context

Abstract

Problem statement: Model Driven Architecture (MDA) is a software development approach based on the design and the transformation of models. In MDA, models are systematically translated to other models and to a source code. Model transformation plays a key role in MDA. Several model transformation languages have been launched lately, aiming to facilitate the translation of input models to output models. The employment of such languages in practical contexts has succeed, although quite often those languages cannot be directly applied to a particular type of model transformation, called refinement. Approach: This study provides a general overview on model refinement and investigates two approaches for model refinement based on Atlas Transformation Language (ATL) referred to as: Refining mode and module superimposition. ATL is a widely adopted language for solving model transformation problems in the MDA approach. Results: This study presents the comparative results obtained from the analysis of the Refining Mode and the Module Superimposition approaches, emphasizing their application benefits. Conclusion: The increasing use of MDA for the design of software systems empowered researches on how developers may benefit from approaches that perform model refinement. The main advantages achieved with the use of the Module Superimposition technique are maintainability and reusability improvement, obtained through module composition and rule superimposition. In its turn, the Refining Mode stands out for its ease of use.