| Title |
From EventStorming artifacts to user stories: a semi-automated requirements extraction approach |
| Authors |
Jurgelaitis, Mantas ; Ramanauskas, Antanas ; Ambrozaitis, Gvidas ; Stulpinaitė, Guoda |
| DOI |
10.1109/ACCESS.2026.3728512 |
| Full Text |
|
| Is Part of |
IEEE Access.. Piscataway, NJ : IEEE. 2026, vol. 14, p. 134943-134956.. ISSN 2169-3536 |
| Keywords [eng] |
EventStorming board ; image processing ; natural language processing ; requirements engineering ; software engineering |
| Abstract [eng] |
Requirements engineering is a critical phase of software development that directly affects project scope, cost, and quality. In software development companies, a requirements list is typically prepared before creating a commercial proposal and signing a contract. EventStorming workshops are widely used for requirements elicitation, as they enable domain experts and developers to collaboratively model business processes using sticky notes. However, although EventStorming is effective at capturing individual requirements within a business process, no established method exists for transforming them into a structured requirements list, such as user stories. Instead, this conversion is performed manually by a business analyst. This paper proposes a methodology that facilitates user story elicitation through a semi-automated transformation of EventStorming workshop artifacts into user stories. The proposed approach was implemented using an SSD MobileNet V1 FPN object detection model for artifact recognition, Google Vision API OCR for text recognition, an artifact clustering algorithm, and a specialized GPT 5.2 LLM agent for requirements generation components. The final output is a project-ready list of user stories. The methodology was validated experimentally, achieving an F1 score of 0.9232 for sticky note recognition and an accuracy of 94.03% for EventStorming classification experiment. Results demonstrate high precision and viability of the approach. |
| Published |
Piscataway, NJ : IEEE |
| Type |
Journal article |
| Language |
English |
| Publication date |
2026 |
| CC license |
|
