Modelltransformasjoner

Modelltransformasjoner (model transformations) are systematic processes for converting models from one form to another within model-driven engineering (MDE). They are defined by metamodels that describe the structure and semantics of both source and target models and by transformation rules that map elements from the source to the target, aiming to preserve meaning where possible.

Types of modelltransformasjoner include model-to-model transformations (M2M), which create a new model from an existing one, and model-to-text transformations (M2T), which generate artifacts such as source code, documentation, or configuration files. Some transformations combine both directions in bidirectional or chained workflows, enabling iterative refinement across model representations.

Core concepts and languages: Metamodels provide the schema for both input and output models; transformation rules encode the correspondences between source and target elements. Transformation approaches can be declarative, specifying what should be produced, or imperative, detailing how to produce it. Notable standards and tools include the OMG’s QVT (Query/View/Transformation), the ATL language, Epsilon, and Acceleo for M2T; graph-transformation techniques such as Henshin are also used in some contexts.

Process and evaluation: Transformations are typically executed by a transformation engine that applies rules to input models, producing output models or artifacts. Key quality attributes include termination, determinism, correctness, and traceability. Validation often uses model test suites, model differencing, and round-tripping tests to ensure consistency between representations.

Applications and impact: Modelltransformasjoner enable automation in software development and engineering, including code generation from high-level models, model synchronization across tools, model refactoring for platform migrations, and support for domain-specific languages. They help improve consistency, productivity, and abstraction level while reducing manual boilerplate.