Enhancing quality of code smell detection on Java byte code
Ghosh, Aritra
A software clone occurs when two code fragments are identical to each other in the source code. Clones are considered harmful for code maintenance and evolution, because it increases complexity of the system; therefore the software maintenance costs. Many code clone detection techniques have been proposed to detect clones like text based, token based, metric based, AST based and PDG based. These techniques have been critically evaluated based on a number of efficiency measure parameters like metric value evaluation, runtime etc. But in the existing methods, many false positive clones are detected which cause low precision value. In this paper, we propose a hybrid approach to improve precision value of clone detection. We have found our hybrid technique better compared to the existing CloneDR in respect to precision, recall, false positive and false negative.