Abstract: Software testing is an important part of software development, which is an important way to guarantee software quality and improve reliability. Testing starts in the requirements analysis phase, while in reality, many requirements are uncertain, and there are many uncertainties in the process of software testing. This paper mainly argues test cases generation methods of software with uncertain requirements, with Partial Model based on the labeled transition systems to model for uncertain requirements, and then to generate effective test cases based on the established model. Aiming at the uncertainty of the model, the probability is introduced, and the probability representation of uncertain parts is presented. Furthermore, the method of test cases generation is proposed based on the characteristics of labeled transition systems and Partial Model. Because the model has uncertain information, the resulting test cases have a certain probability, and the test cases are prioritized based on this.

Key words: software testing, uncertain requirements, test cases, labeled transitions systems, partial models, probability express

摘要： 软件测试是软件开发重要的一部分，是保证软件质量，提高软件可靠性的重要途径。测试在需求分析阶段就开始介入，而在现实中很多需求是不确定的，在软件测试的过程中会面临许多不确定性的挑战。主要研究具有不确定需求的软件测试用例生成方法，利用基于标记迁移系统的Partial Models（部分模型）对不确定的需求进行建模，进而基于建立的模型生成有效的测试用例。针对模型中具有的不确定性引入概率，对不确定的部分进行概率表示，进一步基于标记迁移系统和Partial Models的特点提出测试用例生成方法。由于模型中具有不确定信息，由此生成的测试用例具有一定的概率，基于此对测试用例进行优先级排序。

关键词: 软件测试, 不确定需求, 测试用例, 标记迁移系统, 部分模型（Partial Models）, 概率表示