Detecting Metamodel Evolutions in Repositories of Model-Driven Projects

Model-Driven Engineering (MDE) is a discipline that leverages abstraction and automation in software development. Projects are typically composed of inherently different artifacts, including models, metamodels, model transformations, code generators, and concrete syntax definitions. Despite the increasing availability of reusable projects (e.g., through GitHub), their reuse possibilities depend on the availability of accurate, high-level metadata describing architectural information about the project at hand. In this paper, we enhance an existing approach for extracting relevant architectural information from model-driven projects to detect subsequent metamodel versions in evolution paths. In particular, we are interested in those refactorings that enhance the intrinsic quality of metamodels. The approach has been implemented by extending the existing MDEprofiler tool and has been validated on a dataset consisting of metamodels with different size and covering distinct application domains.