While a model is in general defined as an abstraction of some real world object, in the con-text of software development, models describe products. Indeed, most software development models are used to describe different aspects of that product – separating functionality, from structure, behavior, or code. While these models are separate in their description, they are nonetheless related by manifold dependencies. Changes to a model often cause inconsis-tencies in models that are either directly or indirectly related by these dependencies. Trace-ability relations capture this connection between dependent model elements explicitly and so support stakeholders in coping with change – the inability of doing so is generally recognized as on the most significant reasons for errors and thus project failure.
In preparatory work, the applicant developed a novel approach that helps maintain traceability locally, even as related artifacts change. However, changes in models do not only affect traceability but also other model elements in related models – they cause inconsistencies. The host institution, lead by Prof. Egyed, has intensive experience in understanding, identifying, and resolving inconsis-tencies between related models where inconsistencies convey the existence of errors that span related model elements across multiple models. The aim of the proposed fellowship lays in combining the applicant’s work on understanding the impact of model changes onto traceability with Dr. Egyed’s work on understanding the impact of model changes onto other models – and in doing so, develop an integrated, comprehensive approach combining the use of traceability for reasoning about inconsistencies caused by model changes with the maintenance of traceability facilitated by the knowledge about inconsistencies. More detailed, the proposed project focuses on understanding changes to related software models (e.g., requirements, design, test, and implementation) and on reasoning about how they impact the related models and existing traceability among them. In doing so, the proposed work also integrates existing techniques in the fields of change and inconsistency recognition, trace-ability maintenance, and change propagation. A prototype implementation of the approach is planned and the evaluation of our results will involve industrial or industrial-size examples available at the host institution. It is planned to publish research results in high-profile jour-nals and conferences.