Code Clone Detection on Specialized PDGs with Heuristics

PDG-based code clone detection is suitable for detecting on-contiguous code clones while other detection techniques, line-, token-, or AST-based techniques are not. However, PDG-based detection has lower performance for detecting contiguous code clones than the other techniques. Moreover, PDG-based detection is time consuming, so that application to actual software systems is not feasible. The present paper proposes PDG specializations and detection heuristics for enhancing PDG-based code clone detection. The experiment in this paper shows that the proposed methods are effective for PDG-based code clone detection by applying them to 4 open source systems.