如果我们不使用灰色文献，我们会错过什么证据？
What Evidence We Would Miss If We Do Not Use Grey Literature?
摘要：
背景：在过去的几年里，灰色文献（GL）在软件工程（SE）的二级研究中越来越受到关注。值得注意的是，在传统文献（TL）和GL中寻找证据的多焦点文献综述（MLR）研究特别受益于GL内容的增加。尽管对基于MLR的研究越来越感兴趣，但评估GL如何促进MLR研究的文献仍然很少。目的：本研究旨在评估GL的使用如何有助于MLR研究。通过贡献，我们的意思是，了解GL在多大程度上提供了MLR确实用来回答其研究问题的证据。方法：我们进行了一项三级研究，以确定2017年至2019年期间发表的MLR研究，选择了9项MLR研究。通过定性和定量分析，我们确定了使用的GL，并评估了这些MLR对MLR研究的贡献程度。结果如下：我们的分析发现，1）GL提供了TL中没有的证据，2）大多数GL来源用于提供解决问题的建议，解释主题，并对研究结果进行分类，3）研究中使用了19种不同的GL类型;这些GL主要由SE从业者制作（包括博客文章，幻灯片演示或项目描述）。结论：我们证明了GL如何有助于MLR研究。我们观察到，如果这些GL不包括在MLR中，一些结果将被忽略或削弱。我们还描述了在进行这项调查时所面临的挑战，以及应对这些挑战的潜在方法，这可能有助于未来的SE研究人员。
英文摘要：
Context: Over the last years, Grey Literature (GL) is gaining increasing attention in Secondary Studies in Software Engineering (SE). Notably, Multivocal Literature Review (MLR) studies, that search for evidence in both Traditional Literature (TL) and GL, is particularly benefiting from this raise of GL content. Despite the growing interest in MLR-based studies, the literature assessing how GL has contributed to MLR studies is still scarce. Objective: This research aims to assess how the use of GL contributed to MLR studies. By contributing, we mean, understanding to what extent GL is providing evidence that is indeed used by an MLR to answer its research question. Method: We conducted a tertiary study to identify MLR studies published between 2017 and 2019, selecting nine MLRs studies. Using qualitative and quantitative analysis, we identified the GL used and assessed to what extent these MLRs are contributing to MLR studies. Results: Our analysis identified that 1) GL provided evidence not found in TL, 2) most of the GL sources were used to provide recommendations to solve problems, explain a topic, and classify the findings, and 3) 19 different GL types were used in the studies; these GLs were mainly produced by SE practitioners (including blog posts, slides presentations, or project descriptions). Conclusions: We evidence how GL contributed to MLR studies. We observed that if these GLs were not included in the MLR, several findings would have been omitted or weakened. We also described the challenges involved when conducting this investigation, along with potential ways to deal with them, which may help future SE researchers.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
