Is my model right? Let me ask the expert

Defining a domain model is a costly and error-prone process. It requires that the knowledge possessed by domain experts be suitably captured by modeling experts. Eliciting what is in the domain expert’s mind and expressing it using a modeling language involve substantial human effort. In the process, conceptual errors may be introduced that are hard to detect without a suitable validation methodology. This paper proposes an approach to support such validation, by reducing the knowledge gap that separates modeling experts and domain experts. While our methodology still requires the domain expert’s judgement, it partially automates the validation process by generating a set of yes/no questions from the model. Answers differing from expected ones point to elements in the model which require further consideration and can be used to guide the dialogue between domain experts and modeling experts. Our methodology was implemented as a tool and was applied to a real case study, within the IPERMOB project.