Towards Automated, Formal Verification of Model Transformations

Abstract

Verification of models and model processing programs are inevitable in real-world model-based software development. Model transformation developers are interested in offline verification methods, when only the definition of the model transformation and the metamodels of the source and target languages are used to analyze the properties and no concrete input models are taken into account. Therefore, the results of the analysis hold for each output model not just particular ones, and we have to perform the analysis only once. Most often, formal verification of model transformations is performed manually or the methods can be applied only for a certain transformation or for the analysis of only a certain property. Previous work has presented a formalism to describe the characteristics of model transformations in separate formal expressions called assertions. This description is based on the first-order logic, therefore, if deduction rules are provided, a reasoning system can use an assertion set to automatically derive additional assertions describing additional properties of model transformations. In this paper, we propose deduction rules and present the verification of a model transformation of processing business process models.