期刊名称:International Journal of Hybrid Information Technology
印刷版ISSN:1738-9968
出版年度:2015
卷号:8
期号:11
页码:139-150
DOI:10.14257/ijhit.2015.8.11.11
出版社:SERSC
摘要:Most traditional code clone detections have less accurate results because they ignore the structure of the program itself, and some of them really think about it by creating a complex syntax tree but leading to a high time complexity. Confronting such situation, this paper proposes an efficient metric vector-based code clone detection method using function-calling tree. Considering the two program code to be detected, feature vectors in all defined functions of the two different code are extracted first. Then, two function- calling trees are created according to the function-calling process and node matches each other between two trees, at the same time, the matching similarities are calculated. Finally, by using the bottom-up approach and combining similarity values of all child nodes, the detection can get the similarity of the two program code to be detected. Our experiment selects a set of typical code sample to measure and the results demonstrate that, compared the famous JPlag system, it shows better detection effect.
关键词:code clone; similarity calculating; feature vector; function-calling Tree