Detecting missing requirements in conceptual models

Abstract

Completeness of requirements has persisted as a challenge in software development projects. Particularly, in an iterative and incremental development process, only a small portion of requirements are specified before iterations start. Initial versions of the models may be developed in parallel whilst requirements elicitation is still ongoing. In this paper, we propose a metamodel approach for automating the detection of incompleteness in requirements during the conceptual modeling process. Our approach utilizes the metamodel information to allow the modeler to record explicitly each of the model's "known-unknowns", i.e. items of information which are needed to make the model complete, yet which cannot be inferred from the available requirements.