As the application of model transformation becomes increasingly commonplace, the focus is shifting from model transformation languages to the model transformations themselves. The properties of model transformations, such as scalability, maintainability and reusability, have become important. Composition of model transformations allows for the creation of smaller, maintainable and reusable transformation definitions that together perform a larger transformation. This paper focuses on composition for two rule-based model transformation languages: the ATLAS Transformation Language (ATL) and the QVT Relations language. We propose a composition technique called module superimposition that allows for extending and overriding rules in transformation modules. We provide executable semantics as well as a concise and scalable implementation of module superimposition based on ATL.
A single Model Transformation Chain (MTC) takes a high-level input model rooted in the problem domain and through one or more transformation steps produces a low-level output model rooted in the solution domain. To build a single "almighty" MTC that is in charge of every design, implementation and specific platform concern is a complex task. Instead, we can use several smaller MTCs that are easier to develop and maintain, because each MTC is independently developed focusing on a specific concern. However, the MTCs must interoperate to produce complete applications; this inherently creates dependencies between them, because each MTC generates a part of the final lowlevel model. In this paper, we propose an external and explicit mechanism to track dependencies between the MTCs (i.e., the MTCs are oblivious to the mechanism), which is used to automatically derive correspondence relationships between the final models generated by each MTC. The contribution of our mechanism is the reduction of complexity of building interoperable MTCs because the derived correspondences are resolved after the transformations execution, in the solution domain where the semantics of every concept is welldefined. The resolution process consists of (1) checking the consistency between the models, (2) producing communication bridges or (3) guiding the composition of the models. This paper presents three case studies to illustrate the derivation and resolution of correspondence relationships through the MTCs.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations鈥揷itations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.