Model Transformation By-Example: An Algorithm for Generating Many-to-Many Transformation Rules in Several Model Transformation Languages

Practitioners of model-driven engineering usually find that producing model transformations still requires much effort. The Model Transformation By-Example (MTBE) approach softens this problem by means of the automated creation of model transformations from pairs of source and target model examples. However, the current techniques and tools for MTBE are limited in the sense that they can only produce transformations rules that are limited to just one element in some of the patterns. In addition, some transformation languages cannot directly represent many-to-many transformation rules. To overcome both limitations, this paper presents a MTBE algorithm, that can generate many-to-many rules in several transformation languages even if the transformation language cannot directly represent these kinds of rules. In particular, the implementation of this algorithm with the ATL language is available for practitioners, and it has already generated several MTs that are applied in software development. Finally, the paper compares this algorithm with existing techniques for MTBE.