ABSTRACT
Tunnels represent critical components of transportation infrastructures, where the management of ageing assets requires the integration of structural monitoring, data governance, and advanced digital technologies. In this context, the convergence of Building Information Modelling, ontologies, Virtual Reality, and Large Language Models offers a novel framework to enhance the accessibility, interpretation, and operational use of information during inspection and maintenance activities. The present research explores an integrated workflow aimed at bridging the gap between heterogeneous data sources, ranging from historical documentation to inspection reports and 3D digital models, and the practical needs of inspectors and decision-makers. By combining BIM-based semantic structuring with ontological formalisation, the system enables consistent data interoperability and knowledge representation, ensuring that inspection information can be queried and updated dynamically. Virtual Reality environments are introduced as immersive interfaces to visualise, interact with, and annotate the digital twin of the tunnel, supporting an intuitive understanding of defects and degradation phenomena. Furthermore, the integration of LLM-based conversational agents allows users to access information through natural language, making technical data more accessible even to non-specialist operators. This approach, contributes to the digital transformation of tunnel management practices, promoting transparency, repeatability, and efficiency in the assessment process. The paper discusses the methodological framework, technological implementation, and preliminary outcomes of the research, highlighting how ontology-driven BIM, VR interaction, and AI-based language systems can synergistically improve data exploitation and decision-making in tunnel maintenance.
