Smart requirements modeling: AI-assisted automated generation of UML diagrams from software specifications
Ahmed H. Abase, Hesham Hassan, Fatma A. Omara · Egyptian Informatics Journal · 2026
The proposed intelligent framework combines artificial intelligence with traditional programming to automatically transform natural language software requirements into formal UML representations. It uses a pre-trained spaCy model to extract key entities and relationships, which are then transformed into UML diagrams. The evaluation results demonstrate high accuracy, with precision, recall, and F1-scores averaging 0.92. The framework offers benefits such as formal representation, improved requirement management, enhanced stakeholder communication, and potentially improved software development quality. However, it has limitations in handling complex requirements and is dependent on input quality. This study contributes to software engineering by addressing the challenges of requirement representation and improving development processes.