Open Access

Robin KorfmannORCiD, Patrick BeyersdorfferORCiD, Jürgen MünchORCiD, Marco KuhrmannORCiD

• Natural language processing (NLP) offers the potential to automate quality assurance of software requirement specifications. Especially large-scale projects involving numerous suppliers can benefit from this improvement. However, due to privacy restrictions and domain- and project-specific vocabulary, as such in the aerospace domain, the availability of SRS documents for training NLP tools is severely limited. To provide a sufficient amount of data, we studied algorithms for the augmentation of textual data. Four algorithms have been studied by expanding a given set of requirements from European Space projects generating correct and incorrect requirements. The study yielded data of poor quality due to insufficient accuracy caused by the domain-specific vocabulary, yet, laid the foundation for the algorithms improvement, which, eventually, resulted in an increased set of requirements, which is 20 times the size of the seed set. Finally, an explorative experiment demonstrated the usability of augmented requirements to support AI-based quality assurance.