基于K-means聚类的组合测试用例生成优化算法  

Combinatorial test case optimization based on K-means clustering

作　　者：冯霞[1] 王曙燕[1] 孙家泽[1] 

机构地区：[1]西安邮电大学计算机学院,陕西西安710121

出　　处：《西安邮电大学学报》2015年第1期44-48,共5页Journal of Xi’an University of Posts and Telecommunications

基　　金：国家自然科学基金资助项目(61050003);西安邮电大学青年教师科研基金资助项目(ZL2013-26)

摘　　要：针对组合测试生成的测试用例在程序结构测试中出现冗余的问题,应用K-means聚类算法对基于蚁群算法生成的组合测试用例集进行聚类优化。以白盒测试中的逻辑覆盖为依据,将测试用例程序覆盖差异度作为分类的量化标准,根据测试代价决定聚类数目,在每个聚类簇中抽取处于中心点的测试用例构成新的集合。实验结果表明,该算法可以有效减小测试用例集的规模;对比分析不同覆盖准则,可找到在测试用例标准化过程中最优的逻辑覆盖方法。To eliminate test suite redundancy of combinatorial testing in white-box testing, optimized pairwise test suites are generated based on K-means algorithm merging with ant colony algorithm. While building test suites, the logic coverage strategy in white-box testing is adopted. The difference of program coverage is used as quantitative criteria to classify test cases. The clustering number is decided by test cost. A new test suite is built through clustering system choosing data at clusters＇ center. Simulation experiments show that this method reduces the size of test set effectively. Moreover, this paper presents a method of test standardization by applying different coverage criteria, and thus provides a better solution for test suite data mining technology.

关 键 词：K-MEANS聚类算法 蚁群算法 组合测试 白盒测试 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]