MARS: A metamodel recovery system using grammar inference

Domain-specific modeling (DSM) assists subject matter experts in describing the essential characteristics of a problem in their domain. Various software artifacts can be generated from the defined models, including source code, design documentation, and simulation scripts. The typical approach to DSM involves the construction of a metamodel, from which instances are defined that represent specific configurations of the metamodel entities. Throughout the evolution of a metamodel, repositories of instance models can become orphaned from their defining metamodel. In such a case, instance models that contain important design knowledge cannot be loaded into the modeling tool due to the version changes that have occurred to the metamodel. Within the purview of model-driven software engineering, the ability to recover the design knowledge in a repository of legacy models is needed. A correspondence exists between the domain models that can be instantiated from a metamodel, and the set of programs that can be described by a grammar. In this paper, we propose MARS, a semi-automatic inference-based system for recovering a metamodel that correctly defines the mined instance models through application of grammar inference algorithms. The paper contains a case study that demonstrates the application of MARS, as well as experimental results from the recovery of several metamodels in diverse domains. Key Terms – Metamodeling, grammar inference, generative programming, program