Bidirectional model transformation approaches a comparative study

Bidirectional model transformations (Bx) are considered as one of the cornerstones of model-driven engineering paradigm. Bx re-establishes the consistency between the source and target models, when they are evolved independently. Recently, a large number of approaches for Bx are proposed, each with different capabilities and characteristics. Among them is a novel Bx approach, called EVL+trace, which have previously proposed by the authors. In order to choose an appropriate bidirectional transformation language for a particular task, a developer needs to evaluate and compare Bx approaches. In this paper, a comparison method is proposed in which a family of 19 major criteria, mainly inspired from programming language evaluation criteria and bidirectional features, is used to evaluate Bx languages. In order to investigate the effectiveness of the EVL+trace approach, it is compared to three well-known bidirectional languages including Query/View/Transformation-Relational (QVT-R), Triple Graph Grammar (TGG), and Janus Transformation Language (JTL) based on the elicited criteria. The results of analysis demonstrate the effectiveness of the comparison method for Bx approaches and the superiorities and weaknesses of EVL+trace.