Educational Support for Automated Classification of UML Diagrams Using Machine Learning

As engineering is very much based on modeling, this is also important for education in this field, and teachers sometimes have to check a very large number of models and determine if they are valid or not. In software engineering, for modeling in conformity with the standard Unified Modeling Language, attempts have been made to automatically classify diagrams and determine whether they conform to this language. This paper shows an approach based onmachine learning and possible improvements made using a feature-based dataset, with the objective of more accurately categorizing designated labels. Employing a specialized neural network tailored for feature-based learning, the study endeavors to enhance classification accuracy and efficiency. Comparative analysis against a pre-existing model trained on a diagram images dataset reveals better results in predictive outcomes.