Towards an Automatic Extracting UML Class Diagram from System's Textual Specification
Zakaria Babaalla, Abdeslam Jakimi, Mohamed Oualla, Rachid Saadane, Abdellah Chehri · 2024
Developing a software system from natural language requirements is a complex and delicate task that requires a high level of design and programming expertise. Increasing the level of abstraction used to describe these requirements is the most natural solution. Model-Driven Engineering (MDE) also takes this route, using abstract models as primary entities to generate source code automatically or semi-automatically. Among these models, the UML class diagram occupies a privileged place in object-oriented systems because it not only serves as a basis for communication between developers but also provides a closely aligned static representation of the system implementation. However, creating a UML class diagram from a textual system specification poses a significant challenge due to the inherent imprecision and ambiguity commonly found in natural language expressions. In this paper, we propose a model-centric approach based on deep learning for the automatic extraction of UML class diagrams from textual requirements.