Using a Program Transformation

Domain-Specific Modeling (DSM) allows domain experts to concentrate on the essential characteristics of a problem space without being overwhelmed by the complexities that may occur in the solution space. DSM is focused on the creation of a metamodel for a specific domain, from which instances pertaining to specific configurations of that domain can be constructed. However, as the metamodel undergoes evolutionary changes, repositories of instance models (also called domain models) can become orphaned from their defining metamodel. Within the context of model-driven engineering (MDE), we have developed the Metamodel Recovery System (MARS) which addresses the problem of “metamodel drift” and recovers the design knowledge in a repository of legacy models. MARS is a semi-automatic system that uses grammar inference techniques to recover a metamodel by mining instance models. In addition to the instance models, there are other artifacts that can be investigated in the modeling repository. In this paper we describe an extension to MARS in the form of a type inference capability that is accomplished by the use of a program transformation engine that mines the model compiler code and recovers the type information of fields (or attributes) of metamodel entities.