Search-Based Requirements Traceability Recovery
Search-Based Requirements Traceability Recovery
复制标题
基于搜索的需求可追溯性恢复
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
H. Ammar
中科院分区：
文献类型：
--
作者：
Adnane Ghannem;M. Hamdi;Marouane Kessentini;H. Ammar
一些软件维护任务，如回归测试，要求需求可追溯性应该是最新的。生成需求可跟踪性会消耗大量的时间并且容易出错。目前，大多数可用的工具不提供可追溯性链接的自动恢复。在本文提出的方法中，我们制定的需求可追溯性恢复作为一个搜索问题，使用遗传算法。我们的方法需要作为输入的软件系统和一个需求集，并产生一组之间的系统类和作为输出的输入中引入的要求的痕迹。解决方案包括将每个需求分配给系统的一个或多个类，这些类应该尽可能地最大化需求描述与代码元素名称、API文档和注释之间的文本相似性。三个不同的开源项目的验证结果表明，我们的方法是有效的，正确地生成的需求和类之间的源代码的痕迹与86%的精度和召回率的83%，两者的平均值。
Several software maintenance tasks, such as regression testing, require that the requirements traceability should be up-to-date. Generating requirements traceability consumes a lot of time and is error-prone. Currently most available tools do not offer the automated recovery of traceability links. In the approach presented in this paper, we formulate requirements traceability recovery as a search problem using Genetic Algorithms. Our approach takes as input the software system and a requirement set and produces a set of traces between the classes of the system and the requirements introduced in the input as output. A solution consists of assigning each requirement to one or many classes of the system that should maximize as much as possible the textual similarity between the description of the requirements and the name of code elements, API documentation, and comments. The validation results on three different open source projects show that our approach is effective in correctly generating the traces between the requirements and classes in the source code with a precision of 86% and a recall of 83%, both on average.