A Deep Learning Enhanced UML Framework for Modeling Smart Interactive Museum Ecosystems in Cultural Spaces

Abstract

The digital transformation of cultural institutions requires intelligent, adaptive systems to enhance visitor engagement. Traditional system modeling approaches lack the capacity to integrate real-time learning behaviors and adaptability. This paper presents a novel framework that embeds deep learning modules within Unified Modeling Language (UML) diagrams to design smart museum ecosystems. A convolutional neural network-based classifier, trained on the museum exhibits dataset, is integrated as <<Classifier>> components in UML models. Preprocessing, model training, explainability mechanisms, and system traceability were employed to bridge the gap between design and deployment. The model achieved 91% accuracy and a top-3 accuracy of 96.7%. Expert evaluation confirmed 90% traceability from UML design to implementation. The proposed deep learning-augmented UML framework offers a scalable, interpretable solution for designing next-generation smart cultural spaces, combining artificial intelligence performance with formal system traceability.