Canonical label based grouping in model clone detection

Clones are considered to have negative impacts on software maintenance, and should be removed or detected at least. Model clone detection is a relatively new and important field of clone detection research. ConQAT model clone detection is the state-of-the-art detection tool. However, there are inaccurate clone groups included in its result. Canonical label is an advanced graph isomorphism technique to check the isomorphism of the subgraphs. In this article, we adopt canonical label technology into the grouping phrase of model clone detection, and present an experiment to evaluate the performance. The experimental result shows that overlapping clone groups can be eliminated by canonical label based grouping.