Proposal of a similarity measure for unified modeling language class diagram images using convolutional neural network

Jebli, Rhaydae
El Bouhdidi, Jaber
Yassin Chkouri, Mohamed

International Journal of Electrical and Computer Engineering
Vol 14, No 2: April 2024

The unified modeling language (UML) represents an essential tool for modeling and visualizing software systems. UML diagrams provide a graphical representation of a system's components. Comparing and processing these diagrams, for instance, can be complicated, especially as software projects grow in size and complexity. In such contexts, deep learning techniques have emerged as a promising solution for solving complex problems. One of these crucial problems is the measurement of similarity between images, making it possible to compare and calculate the differences between two given diagrams. The present work intends to build a method for calculating the degree of similarity between two UML class diagrams. With a goal to provide teachers a helpful tool for assessing students' UML class diagrams.