Integrating Indoor Localization, Personalization, and Recommendation To Enhance Museum Visitor Experiences

Museums are increasingly adopting innovative strategies to engage visitors and increase attendance. In this context, this thesis proposes a novel approach to enhance the user experience. It first introduces a framework that integrates three digital technologies: Indoor Localization, Personalization, and Recommendation, and then explores each component in depth. We successfully created an Indoor Localization system for Palazzo Barberini, part of Rome's prestigious Gallerie Nazionali di Arte Antica. For personalization, we investigated how Large Language Models (LLMs) can tailor audio guides to individual visitor categories. Our findings suggest that personalized descriptions can be more effective in meeting user needs for most visitor categories. We also tested LLMs for question answering and generation, demonstrating their potential to support personalized interactions between visitors and artworks. We investigated dwell time (i.e., how much time visitors spend viewing each artwork) as a proxy for preference. This led us to study also \textit{museum fatigue}, the phenomenon where visitors gradually spend less time at artworks as their visit progresses. Our results confirm that dwell time connects to artwork likability and provide empirical evidence that museum fatigue exists. Ultimately, we study how to integrate dwell time in the recommendation process. While some results underline its potential, further validation is still required. This thesis makes novel contributions across multiple fields, demonstrating how the combination of interdisciplinary practices can address challenges in the museum sector. It establishes a strong foundation for developing and implementing the proposed framework, bridging computer science, human-computer interaction, and museology.