Effective and efficient model clone detection

Code clones are a major source of software defects. Thus, it is likely that model clones (i.e., duplicate fragments of models) have a significant negative impact on model quality, and thus, on any software created based on those models, irrespective of whether the software is generated fully automatically (“MDD-style”) or hand-crafted following the blueprint defined by the model (“MBSD-style”).

Unfortunately, however, model clones are much less well studied than code clones. In this paper, we present a clone detection algorithm for UML domain models. Our approach covers a much greater variety of model types than existing approaches while providing high clone detection rates at high speed.