The results of the requirements engineering process are predominantly documented in natural language requirements specifications. Besides the actual requirements, these documents contain additional content such as explanations, summaries, and figures. For the later use of requirements specifications, it is important to be able to differentiate between legally relevant requirements and other auxiliary content. Therefore, one of our industry partners demands the requirements engineers to manually label each content element of a requirements specification as "requirement" or "information". However, this manual labeling task is time-consuming and error-prone. In this paper, we present an approach to automatically classify content elements of a natural language requirements specification as "requirement" or "information". Our approach uses convolutional neural networks. In an initial evaluation on a real-world automotive requirements specification, our approach was able to detect requirements with a precision of 0.73 and a recall of 0.89. The approach increases the quality of requirements specifications in the sense that it discriminates important content for following activities (e.g., which parts of the specification do I need to test?).
Non-functional requirements (NFRs) are commonly distinguished from functional requirements by differentiating how the system shall do something in contrast to what the system shall do. This distinction is not only prevalent in research, but also influences how requirements are handled in practice. NFRs are usually documented separately from functional requirements, without quantitative measures, and with relatively vague descriptions. As a result, they remain difficult to analyze and test. Several authors argue, however, that many so-called NFRs actually describe behavioral properties and may be treated the same way as functional requirements. In this paper, we empirically investigate this point of view and aim to increase our understanding on the nature of NFRs addressing system properties. We report on the classification of 530 NFRs extracted from 11 industrial requirements specifications and analyze to which extent these NFRs describe system behavior. Our results suggest that most "non-functional" requirements are not non-functional as they describe behavior of a system. Consequently, we argue that many so-called NFRs can be handled similarly to functional requirements.Comment: Proceedings of the 38th International Conference on Software Engineering, 201
20 October 2022 in Journal of Systems and Software