An Extended MDA Architecture for Ensuring Semantics-Preserving Transformations

MDA is a current trend that shifts the focus and effort from implementation during the development process to models, metamodels and transformations. The four-layer MDA architecture still suffers from the lack of formalization of its artifacts and processes leading to ambiguity and low reliability problems. For instance, currently there is no way for ensuring that MDA transformations are semantic-preserving. In this paper we tackle this problem by incorporating well-consolidated formal methods approaches in the MDA architecture. We propose extending the M1 and M2 layers of the MDA architecture towards the definition of static and dynamic semantic models, both crucial to extract the semantics of any model. The extension includes: (i) a semantic metamodel and a set of transformations representing semantic equations; and (ii) a set of evaluation reduction rules for semantic models in order to extract the meaning from their states and automatic proofs of conformance between these models. In order to validate this work, a proof of concept has been implemented and evaluated verifying conformance between models that represent imperative constructs of programming languages in the modelware context.