Augmenting the Problem Frames Approach with Explicit Data Descriptions Using ChatGPT

Problem Frames (PF) as a mainstream requirements analysis method has attracted extensive attention and research in the field of requirements engineering. However, the current problem diagram as a PF visualization tool is more abstract and lacks the ability to express data descriptions. In this paper, we propose a method to extend problem diagrams based on ChatGPT. The aim is to detail the requirements to ensure a more accurate and comprehensive representation of the complex requirements of the system. We combine problem diagrams and ChatGPT by letting them interact with each other, and obtain the cause table describing causal relationships that exist within shared phenomena, then we model the cause table and calculate the structural coefficient of each cause and rank the causes from the largest to the smallest. Finally the ranked causes are used as an extension of the shared phenomenon, thus using the property that they can help improve the detailed descriptions of the requirements. A case study of an insulin injection control system demonstrates the applicability and viability of this method.