A New Approach for Bidirectional Model Transformation Based on the Epsilon Framework and Traceability Techniques

Model-driven development is a new paradigm in software development in which the implementation code is automaticallysemi-automatically generated from the model, using a set of techniques named model transformations. All activities in model-driven approaches are performed via the application of transformations. One of such activities is model-to-model transformation, which defines a unidirectional relation between the source and target models in the simplest case studies. In this case, the only possibility is to produce the target model from the source model. However, in general, the source and target models are evolved independently. Therefore, a more advanced transformation called bidirectional transformation (Bx) is required to restore consistency. Bx approaches have several challanges such as ambiguities in transformation language, special notation, and expensive formalization. In order to cope with such challenges, in this paper, a new Bx approach is proposed based on the Epsilon model-driven framework and traceability techniques. In this approach, the relation between the source and target models is formalized, and then is implemented on a well-known benchmark. The proposed Bx approach is compared with three bidirectional approaches based on ten criteria. The result of comparison demonstrates the superiority of the proposed approach in terms of portability, preservation, reconciliation, and operation support.