Detecting Model Clones using Design Metrics

Abstract

The cloning in software is a frequent phenomenon that leaves a negative impact among the product lines or the version control where the developer is involved in the evolution of software system due to any enhancement or changing requirements that leads to a release of new version. With the advent of MDD, identification of clones shifted from code to models to tackle risks at early stages. Due to the renaming of model elements, some model clones are missed that reports secondary clones instead of primary. So, in order to increase recall of clones in models we have proposed a hybrid approach based on the tree, lexical and metric approaches and validated it using SDMetric tool followed by analysis of detection of exact, renamed clones and modified clones. It provides one to one mapping in the form of corresponding primary clones with maximal matching that helps to reduce the domain for comparison of code at the implementation level. Such clones need to be identified among the versions and the stable part with least changes acts as a pattern and can be reused by the product lines.