基于演化模式特征的克隆代码分类  

Classification for code clones based on evolution pattern feature

作　　者：王春晖[1] 张久杰[2] 刘志国[1] 张丽萍[1] 刘东升[1] WANG Chun-hui ZHANG Jiu-jie LIU Zhi-guo ZHANG Li-ping LIU Dong-sheng(Computer and Information Engineering College, Inner Mongolia Normal University, Hohhot 010022, China Department of Computer, Baotou Teacher＇s College, Inner Mongolia University of Science and Technology, Baotou 014000, China)

机构地区：[1]内蒙古师范大学计算机与信息工程学院,内蒙古呼和浩特010022 [2]内蒙古科技大学包头师范学院信息科学与技术学院,内蒙古包头014000

出　　处：《计算机工程与设计》2017年第8期2121-2126,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61363017;61462071);内蒙古自然科学基金项目(2016MS0613)

摘　　要：克隆代码稳定性与它在多版本演化过程中的变化模式相关,综合这些变化模式并构建自动分类模型,实现克隆代码稳定与易变的特征标注,便于开展克隆分类方面的研究。从克隆演变、克隆规模变化与克隆修改3个维度将克隆演化分为12种演化模式,通过版本间克隆的差异分析识别这12种模式,记录变化参数,使用克隆聚类特征模型计算每个克隆实例的特征向量,用聚类算法实现克隆稳定性分类。对两款软件进行实验分析,分析结果表明,基于该方法可以得到关于克隆稳定性的有效分类数据集。The clonal stability is related to its change pattern in the process of multi-version evolution. Combining these change patterns, the automatic classification model was constructed and marked by stable and variable feature, which was convenient for some research related to clonal classification. The clonal evolution was divided into 12 evolution modes, and the 12 patterns were identified through the difference analysis between clones. The change parameters were recorded. The feature vector of each clone instance was calculated using clonal clustering feature model. Clonal stability classification was realized using clustering algorithm. Results of experiments on two multi-version softwares show that the proposed method can obtain the effective classification data set for clonal stability.

关 键 词：克隆代码 克隆演化模式特征 克隆分类 克隆分析 克隆管理 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]