A formal graph transformation based language for model-to-model transformations
Aditya Agrawal, Gábor Karsai · 2004
starting from machine code to languages for component-oriented systems. The next step is towards developing Domain-Specific Languages (DSLs) for particular domains. Frameworks can be used to facilitate the development of such DSLs. However, current frameworks often lack a formal methodology for specifying the dynamic semantics. Dynamic semantics can be specified as a mapping from the DSL to a known model-of-computation. In this dissertation it is argued that graph grammar and transformation theory can be used to develop a language for the specification and automated implementation of model-to-model transformations. Model-to-model transformations introduce new challenges: (1) multiple graph domains may be involved in the transformation, (2) tight control over the execution of transformations is required, and (3) efficiency of the transformations is a major concern. Graph Rewriting and Transformations (GReAT) is a language developed to prove the research hypothesis and to address these challenges. Key features of the research are the development of methodologies for transformations across domains, language design, and the creation of efficient transformation algorithms. These algorithms were used to develop a virtual machine for GReAT called the Graph Rewriting Engine (GRE). A case study is presented that demonstrates the usefulness of GReAT in solving real world problems. Another case study presents a proof of correctness of a transformation as an example of the formal verification of GReAT transformations.