Refinement equivalence in model-based reuse: Overcoming differences in abstraction level

Reuse of models assists in constructing a new model on the basis of existing knowledge by retrieving a model that matches a preliminary partial input model or some facts about the domain and adapting it to the current needs. It often employs similarity measures to identify reusable models that are structurally and semantically similar to the input model. However, in many cases, an input model, being a preliminary one, is of a higher level of abstraction than the detailed models to be retrieved, and structural similarity cannot be detected. This paper proposes the concept of refinement equivalence, which means that a detailed model is a refinement of an abstract input model. Measuring refinement equivalence rather than structural similarity enables retrieving an appropriate model, despite differences in the abstraction level between the models. The paper discusses the structural characteristics of refinement operations in Object-Process Methodology (OPM) models and presents an algorithm that detects refinement equivalence.