Automated support to capture verbal just-in-time requirements via audio mining and cluster-based visualization
Published on Jul 1, 2018in Journal of Industrial Information Integration
· DOI :10.1016/j.jii.2018.06.001
Abstract Being able to quickly and accurately capture requirements is crucial when using agile methodologies. Requirements, to that end, are often captured in an as-needed and informal manner, with continuous stakeholder interaction. Techniques such as interviews, user stories, rapid feedback, and text mining are commonly used in the industry to gather such informal requirements, which are often congruent with the concept of ‘just-in-time’ (JIT)—informally represented and continuously refined—requirements. Oral communication is an integral part of most of these techniques, and the loss or misinterpretation of verbally communicated requirements is a daunting problem. To address this issue, In this invited paper, we propose a novel framework to assist developers in capturing verbal requirements in complete manner. In particular, our framework automatically captures, transcribes, and mines verbal communication, which in turn produces a set of loosely formulated candidate requirements for further elaboration. We also conducted a survey among practitioners concerning requirements in agile methodologies and our framework. The results of this survey provide positive feedback on the prospects of the framework and also indicate the prevalence of JIT requirements in closed source agile projects. In addition, we proposed a new framework for visual exploration of JIT requirements and presented our JITREvisu prototype tool.