Requirements Refinement Using Domain knowledge

Abstract

Getting clear and accurate requirements specifications is still a challenging task for requirements analysts for two main reasons: i) Despite the fact that analysts are experts in software technology, their knowledge in the domain area where the system is developed is often limited, thereby limiting the effectiveness of requirements elicitation; ii) Most of the requirements in agile development methods are captured using natural language, which is vulnerable to different interpretations and understanding leading to inaccurate requirements specifications. Although there are some methods that support the analyst in producing highquality requirement specifications, a small number of researchers focus on agile requirements engineering with user stories. However, their approaches are either manual or do not use any kind of domain knowledge as a reference to support the analysts in refining the requirements. This thesis has been performed to identify the potential use of domain knowledge in order to help the analyst to deliver high-quality requirement specifications. We developed the Story Suggestor Tool, which uses business process models as source of domain knowledge that can be utilized to suggest new requirements to the analyst. The development of the artifact has been guided by design science methodology. To validate our approach, an experiment has been conducted to check whether or not our developed artifact, the Story Suggestor Tool, helps the analyst in delivering higher-quality requirements. Based on the evaluation of the experiment results, we concluded that the tool helps the analyst in producing relatively more complete and correct requirements.