基于改进LDA算法的测试用例优先级排序研究

Research on Test Case Prioritization Algorithm Based on Enhanced LDA

提出了一种新的静态黑盒测试用例优先级(Test Case Prioritization,TCP)排序算法,在没有源代码的情况下,需要对测试用例集合按照优先级进行排序。改进原始的LDA算法,应用每个测试用例的语言数据(如,标识符名称、注释和字符串文字)和标签来预测测试用例的主题,利用期望最大化算法进行模型的参数估计。在实验部分,我们将提出的算法与现有的技术进行比较,以验证该算法的性能。

This paper presents a new static black box Test Case Prioritization （TCP） sorting algorithm. In the absence of source codes, to prioritize the collection of test cases is needed. This paper improves the original LDA algorithm, and uses the linguistic data of each test case （i.e., identifier, comments and string literals） and tags to predict topic of test case, using the expectation maximization algorithm to estimate the parameters of the model. In the experimental part, we compare the proposed algorithm with existing technologies to verify performance of the proposed algorithm.