多准则优化的规模约束型测试用例选择  

Multiple criteria optimization for scale constrained test case selection

作　　者：吴先平[1] 

机构地区：[1]复旦大学附属上海市第五人民医院,上海200000

出　　处：《电子设计工程》2016年第24期49-52,56,共5页Electronic Design Engineering

摘　　要：软件修改之后可以重新测试之前的所有用例来发现错误,但是这种方法耗费巨大,为了减少测试用例数量,优化测试工作,本文提出了一种全新的用例选择方法,即从现有的测试用例集中挑选一定数量的用例并进行重新排序。该方法塑造了一个线性规划问题,采用两个代码覆盖准则并放宽约束来发现接近最优方案的用例,然后对这些用例使用投票机制获得最优用例集,最后采用最大化最小覆盖的贪心算法进行迭代排序。实验表明在大部分案例中,新方法的性能相比现有方法有显著的改进,而且一致性更好。One traditional approach to detect errors of software after modification is to rerun all the previous test cases, which is too costly. To optimize test and reduce test cost, we proposed a new approach which is to select a predefined quantity of test cases from existed cases and re-order them. This approach forms an IP problem, uses two coverage-based criteria and constraint relaxation to find cases close to best solution, then a voting mechanism is used to select the subset cases and then are prioritized with Maximize minimum coverage based Greedy algorithm in regression. The proposed approach was evaluated against other existing ones with experiment and the result showed that the new approach performed better in most cases with higher consistency.

关 键 词：软件回归测试 测试用例选择 线性规划 

分 类 号：TN99[电子电信—信号与信息处理]