Many test case prioritization (TCP) studies based on regression testing using a code-based development approach have appeared. However, few studies on model-based mutation testing have explored what kind of fault seeding is appropriate or how much the code-based results differ. In this paper, as automatic seeding for the mutation generation, several mutation operators were employed for the UML statechart. Here, we suggest mutation testing employing the model-based development approach and a new TCP method based on an alternating variable method (AVM). We statistically compare the average percentage of the fault detection (APFD) results of the new method to other TCP methods such as a greedy algorithm for code coverage or fault exposure possibility. Finally, in empirical studies, the model-based TCP results for a power window switch module, a body control module, and a passive entry and start system are derived; these are real industrial challenges in the automotive industry.

中文翻译：

---

使用交替变量方法的基于模型的测试案例优先级用于基于UML的模型的回归测试

基于基于代码的开发方法的回归测试，许多测试案例优先级划分（TCP）研究已经出现。但是，很少有基于模型的突变测试的研究探索哪种故障播种是合适的，或者基于代码的结果有多少不同。在本文中，作为自动生成突变的种子，UML状态图使用了几种突变算子。在这里，我们建议使用基于模型的开发方法和基于交替变量方法（AVM）的新TCP方法进行突变测试。我们将新方法的故障检测（APFD）结果的平均百分比与其他TCP方法（例如用于代码覆盖率或故障暴露可能性的贪婪算法）进行统计比较。最后，在实证研究中，电动窗开关模块基于模型的TCP结果，导出车身控制模块和被动进入与启动系统。这些是汽车行业真正的工业挑战。