A Framework for Reviewing and Improving Non-Functional Requirements in Agile-based Requirements

The ill-definition or neglect of non-functional requirements has been frequently reported as a problem in agile requirements engineering. Despite the importance of such requirements for the success of software projects, little research is available for reviewing them in the agile requirements. The few existing proposals are either too heavy-wight, lack agility assessment, or lack clarity on how to be embedded in an agile environment. These factors might reduce their adoption among agile practitioners. To address this problem, we propose a framework for reviewing non-functional requirements in the agile requirements formulated as user stories. Our proposed framework utilizes a combination of natural language processing and artificial intelligence techniques to automate analyzing and predicting non-functional requirements in user stories. Then, the artifacts needed for the requirements reading and reviewing technique are generated to assist requirements inspectors in reviewing non-functional requirements, identifying possible defects, and consequentially making a reasonably informed review decision. Our proposal aims to mitigate the most reported causes of the problem in the literature: the minimal documentation, the functionality-driven nature of agile requirements, and insufficient non-functional requirements knowledge.