This is a research project targeted at the generation of model-to-model transformations based on examples.
The generator requires the following input:
- Meta models of source and target, i.e.
- Meta model instances which correlate, i.e.
The output is a model-to-model transformation from the source to the target meta model in [ETL](TODO LINK Epsilon Transformation Language).
-
Requirements:
- Java 8
- Eclipse Epsilon
- Microsoft SQL Server 2012 (Install using the instance Name: MSSQLSERVER11 instead of DEFAULT)
-
Quick Start:
- import /src/ into your Eclipse workspace
- import /ext/eclipse.userlibraries into Eclipse
- create the MS SQL database using /src/TransformationGeneratorEvaluation-Database.sql
- execute /src/TransformationGenerator/src/transformationGenerator/main/Starter.java: The best solution(s) found will be shown in the output window at the end.
- evaluate execution using /src/TransformationGeneratorEvaluation.xlsx: shows fitness convergence and provides an overview of the fitness landscape.
The quality of the transformation depends on the examples. Additionally, the generator is limited to the creation of a sub-set of all transformations. Thus, the solution might be incomplete and must be refined manually. For more information please read the publications:
- Automatic Generation of Model-to-Model Transformations using Evolutionary and Swarm-Intelligent Algorithms Thomas Kühne
- Patterns for Constructing Mutation Operators: Limiting the Search Space in a Software Engineering Application Thomas Kühne, Heiko Hamann, Svetlana Arifulina and Gregor Engels; LNCS 9594, p. 278 ff.