What have we learned from the challenges of (semi-) automated requirements traceability? A discussion on blockchain applicability

<br>Over the last three decades, researchers have attempted to shed light into the requirements traceability problem by introducing tracing tools, techniques, and methods with the vision of achieving ubiquitous traceability. However, despite the technological advances, requirements traceability remains problematic for researchers and practitioners. In this scenario, we aim to identify and investigate the main challenges in implementing (semi-)automated requirements traceability, as reported by literature. We carried out a systematic literature review based on the guidelines for systematic literature reviews in software engineering, proposed by Kitchenham. We retrieved 4530 studies by searching five major bibliographic databases and selected 70 primary studies. These studies were analyzed and classified according to the challenges they present and/or address. The systematic literature review revealed 21 challenges of (semi-)automated requirements traceability which led to the derivation of a set of requirements. To address these requirements, blockchain technology was proposed as a suitable solution for software and systems traceability across organizational boundaries. Findings reveal that the most frequently presented and/or addressed challenges are technological challenges, in particular low accuracy of traceability recovery methods. Findings also suggest that further research is needed, especially through case studies or surveys, to explore traceability practices in industrial settings. Finally, based on our findings we recommend researchers to leverage the human facet of tracing, distributed traceability and traceability approaches to support agile and DevOps practices.

近三十年来，研究者们致力于破解需求可追溯性（requirements traceability）难题，通过引入各类追溯工具、技术与方法，以期实现泛在可追溯性（ubiquitous traceability）。然而，尽管技术持续迭代，需求可追溯性仍是研究者与从业者面临的棘手问题。
在此背景下，本研究旨在识别并调研文献中提及的半自动化与自动化需求可追溯性（(semi-)automated requirements traceability）落地面临的主要挑战。本研究遵循Kitchenham提出的软件工程系统文献综述（systematic literature review）指南，开展了系统性文献综述工作：通过检索五大主流文献数据库，共获取4530篇相关研究，并从中筛选出70篇原始研究。随后，我们依据这些研究提出或探讨的各类挑战，对其进行了分析与分类。
本次系统文献综述共识别出21项半自动化与自动化需求可追溯性相关挑战，并由此推导得到一组需求规范。针对上述需求，区块链（blockchain）技术被提出作为跨组织边界的软件与系统可追溯性的适配解决方案。
研究结果显示，最受关注且被广泛探讨的挑战为技术类挑战，尤以追溯恢复方法（traceability recovery methods）的准确率偏低最为突出。研究结果同时表明，仍需开展进一步研究——尤其是通过案例研究与调研，以探索工业场景中的可追溯实践。最后，基于本次研究结果，我们建议研究者借助追溯的人为因素、分布式可追溯性及相关可追溯方法，以支撑敏捷实践与开发运维一体化（DevOps）实践。