Clone Detection Using Time Series and Dynamic Time Warping Techniques

This paper presents a novel approach to detect code clones. The proposed approach formulates the clone detection problem as a problem of querying and mining time series data [18]. The approach is composed of three steps. The first step extracts modules (i.e., methods, functions...) from the software system, the second transforms modules to time series and the third one calculates the similarity degree between modules using the DTW (i.e., Dynamic Time Warping) algorithm. Two modules are reported as clones if the DTW similarity value between them is greater than some predefined threshold. The results of the experiment conducted on well known software systems shown that our approach has the potential to detect clones of type I, type II and type III in an effective manner.