基于模板导向的遗传规划软件自动修复方法

Automatic Fixing Based on Template-Oriented Genetic Programming

随着互联网的飞速演变,技术不断革新和迭代,应用程序变得越来越臃肿和复杂,程序维护成本也成倍增加.面对多次迭代的程序,传统的随机选择修复方法在样本空间中修复问题,但这样修复的样本过大,效率并不高.提出一种基于带模板导向的遗传规划的软件自动修复方法,以降低维护成本.通过分析抽象语法树的控制流图信息,从程序结构中获取特征信息,作为考察程序匹配程度的两个维度,选出最适合的模板程序.然后使用匹配后的模板程序信息进行错误定位和遗传规划修复.在使用网络公开的数据集Tech-V3.5.2进行对比实验后,结果表明,该算法在软件修复的准确性和时间上均有提高.

syntax tree,the feature information is obtained from the program structure,which is used as two dimensions to examine the degree of program matching,and the most suitable template program is selected.The matched template program information is then used for error localization and genetic planning fixes.After comparative experiments using Tech-V3.5.2,the results show that the algorithm proposed in this paper improves the accuracy and time of software repair.