Defining a New Digital Twin Ontology for Cultural Heritage Preservation – the Case of ARGUS

DIGITAL HERITAGE (2025) S. Campana, D. Ferdani, H. Graf, G. Guidi, Z. Hegarty, S. Pescarin, and F. Remondino (Editors) Defining a New Digital Twin Ontology for Cultural Heritage Preservation – the Case of ARGUS G. Pavlidis , V. Sevetlidis and V. Arampatzakis Athena Research Center, Greece Abstract The sustainable preservation of cultural heritage (CH) assets increasingly demands predictive monitoring approaches that integrate multimodal data and decision support mechanisms. EU project ARGUS introduces a semantic ontology designed to unify sensor observations, diagnostic activities, risk predictions, and conservation decisions within a coherent, operational framework. Building upon standards such as CIDOC CRM, SOSA/SSN, PROV-O, GeoSPARQL, and OWL-Time, the ontology advances heritage computing toward dynamic condition monitoring and preventive conservation strategies. CCS Concepts •Information systems →Ontology modeling; •Theory of computation →Knowledge representation and reasoning; 1. Introduction Cultural heritage (CH) assets are increasingly at risk from environmental, anthropogenic, and structural factors. Effective preservation demands proactive strategies based not only on static documentation, but on dynamic, multimodal, and predictive monitoring approaches. While established ontologies such as CIDOC CRM [Doe03] provide robust models for static heritage descriptions, they lack constructs for real-time condition tracking, risk forecasting, and decision support. The ARGUS project (Horizon Europe, Grant Agreement No. 101132308) addresses this gap by developing an integrated framework for remote monitoring, digital twin creation, and predictive management of cultural heritage sites. Within ARGUS, we introduce PANOPTES, a new digital twin ontology designed to semantically integrate sensor observations, diagnostic processes, risk predictions, and conservation decisions into a unified, operational model. PANOPTES builds upon standards such as SOSA/SSN [JHC∗19] (sensor observations), PROV-O [MM13] (provenance tracking), GeoSPARQL [PH12] (spatial modeling), and OWL-Time [HP06] (temporal representation), while maintaining compatibility with CIDOC CRM principles for cultural heritage description. Recent work on semantic models for heritage digital twins [FN25] highlights the critical need for predictive, dynamic, and multimodal representations, further motivating the development of integrated frameworks such as PANOPTES. Through this integration, PANOPTES enables the evolution of cultural heritage management from reactive documentation to predictive, evidence-based conservation planning. 2. PANOPTES Ontology Overview PANOPTES models the complete dynamic preservation workflow, structured around the following core entities: •Asset: A cultural heritage entity, modeled as a specialization of sosa:FeatureOfInterest and geo:Feature. •Measurement: A specialization of sosa:Observation, capturing spatiotemporal data about asset condition. •Instrument: Devices or methods (sosa:Sensor) used to perform Measurements. •Diagnosis: An interpretation of Measurement data, modeled as aprov:Activity, assessing asset conditions. •Prediction: A forecasted outcome (prov:Entity) derived from computational models based on Diagnoses. •Threat: A risk factor potentially impacting an Asset, semantically linked to spatial and temporal contexts. •Decision: An action planning outcome, governed by Policies and Rules, to mitigate predicted risks. •Cultural Documentation: Structured records of observations, diagnoses, and interventions, following cidoccrm:E31_Document. Spatial properties are represented using GeoSPARQL [PH12], while temporal evolution is modeled using OWL-Time [HP06]. Provenance tracking of all observations, computations, and decisions is ensured through PROV-O [MM13]. 2.1 Ontological Relations PANOPTES defines a rich set of semantic relations among its core entities, structuring the dynamic monitoring and decision-making process: © 2025 The Author(s). Proceedings published by Eurographics - The European Association for Computer Graphics. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. DOI: 10.2312/dh.20253260 https://diglib.eg.org https://www.eg.org 2 of 2 G. Pavlidis & V. Sevetlidis & V. Arampatzakis / Defining a New Digital Twin Ontology for Cultural Heritage Preservation – the Case of ARGUS •Asset is observed through one or more Measurements, each linked to specific Instruments and executed following a defined Protocol. •Measurement produces a Quantity result and is characterized by spatial (geo:Geometry) and temporal (time:Instant) attributes. •Measurement data are interpreted via a Diagnosis, representing an analytic activity that assesses the Asset’s condition over time. •Diagnosis is generated using a Computational Model, which encapsulates methods for condition inference or deterioration assessment. •Diagnosis informs a Prediction, modeling expected future states or risks impacting the Asset. •Prediction is linked to specific Threats, which are categorized according to their origin (e.g., environmental, structural, anthropogenic). •Threats are associated with potential Events (e.g., structural failure, environmental stress episodes) that may require urgent intervention. •Decision entities are informed by Diagnoses and Predictions, following Policies and detailed Rules that regulate conservation strategies. •Policy defines general conservation goals, while Rules specify operational thresholds, actions, or mitigation procedures. •Visualization entities are linked to Computational Models and Predictions to facilitate stakeholder understanding through graphical interfaces. •Cultural Documentation maintains provenance records, linking Observations, Diagnoses, Predictions, Decisions, and conservation Events back to their originating Assets. •Each Event can retroactively update the Asset’s state and trigger new Measurements, closing the monitoring-feedback loop. Through these relations, PANOPTES creates a dynamic semantic graph capturing not only the static features of cultural heritage assets but also their evolving conditions, inferred risks, and managed interventions over time. 2.2 Design Principles PANOPTES is founded on the following design principles: •Asset-Centric Modeling: All activities are linked to specific cultural assets. •Dynamic State Representation: Support for temporal evolution and condition monitoring. •Predictive Maintenance Integration: Formal representation of forecasts and interventions. •Semantic Interoperability: Alignment with CIDOC CRM, SOSA/SSN [JHC∗19], PROV-O, GeoSPARQL, and OWL-Time standards. •Operational Readiness: Natural mapping to a relational database schema for scalable deployment. 3. Pilot Deployment within ARGUS PANOPTES is piloted within the European Union’s ARGUS project (Grant Agreement No. 101132308), focusing on proactive conservation of at-risk heritage sites. Two primary pilot scenarios are under development: •Structural Monitoring: Continuous acquisition of structural deformation measurements (e.g., crack width monitoring) for historical monuments. Measurements are semantically linked to assets, diagnoses infer deterioration trends, and predictions forecast risk escalation. •Environmental Risk Monitoring: Integration of environmental sensor data (temperature, humidity, pollutants) inside museums and historical structures. Diagnoses detect unfavorable microclimatic conditions, while predictions guide preventive interventions to mitigate material degradation. PANOPTES supports real-time data ingestion, dynamic updates of asset states, and traceable decision-making workflows aimed at minimizing deterioration risks. 4. Conclusion and Future Work PANOPTES advances the state of the art in cultural heritage management by bridging static documentation models with predictive monitoring and decision support. Through its integration of multimodal data streams, predictive analytics, and conservation planning into a coherent semantic framework, it enables a shift from reactive to proactive heritage preservation strategies. Future work will focus on scaling deployment across diverse heritage asset types, extending semantic models to support AI-driven risk assessment techniques, and enhancing visualization tools to facilitate stakeholder interaction with digital twins. Acknowledgments This work has received funding from the European Union’s Horizon Europe programme under grant agreement No. 101132308 (ARGUS project). References [Doe03] DOERR M.: The cidoc conceptual reference model: An ontological approach to semantic interoperability of metadata. AI Magazine 24, 3 (2003), 75–92. 1 [FN25] FELICETTI A., NICCOLUCCI F.: Artificial intelligence and ontologies for the management of heritage digital twins data. Data 10, 1 (2025), 1–23. 1 [HP06] HOBBS J. R., PAN F.: Time ontology in owl. https://www. w3.org/TR/owl-time/, 2006. W3C Working Draft. 1 [JHC∗19] JANOWICZ K., HALLER A., COX S., PHUOC D. L., TAYLOR K.: Sosa: A lightweight ontology for sensors, observations, samples, and actuators. Journal of Web Semantics 56 (2019), 1–10. 1,2 [MM13] MOREAU L., MISSIER P.: Prov-o: The prov ontology. https: //www.w3.org/TR/prov-o/, 2013. W3C Recommendation. 1 [PH12] PERRY M., HERRING J.: Ogc geosparql – a geographic query language for rdf data. https://www.ogc.org/standards/ geosparql, 2012. OGC Standard. 1 © 2025 The Author(s). Proceedings published by Eurographics - The European Association for Computer Graphics.