Towards an approach for evaluating the quality of requirements

In engineering design, the needs of stakeholders are often captured and expressed in natural language (NL). While this facilitates such tasks as sharing information with nonspecialists, there are several associated problems including ambiguity, incompleteness, understandability, and testability. Traditionally, these issues were managed through tedious procedures such as reading requirements documents and looking for errors, but new approaches are being developed to assist designers in collecting, analysing, and clarifying requirements. The quality of the end-product is strongly related to the clarity of requirements and, thus, requirements should be managed carefully. This paper proposes to combine diverse requirements quality measures found from literature. These metrics are coherently integrated in a single software tool. This paper also proposes a new metric for clustering requirements based on their similarity to increase the quality of requirement model. The proposed methodology is tested on a case study and results show that this tool provides designers with insight on the quality of individual requirements as well as with a holistic assessment of the entire set of requirements.