Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation

Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation George Pavlidis1, Anestis Koutsoudis1, Despoina Tsiafakis1, Melpomeni Karta1, Vasileios Sevetlidis1, Vasileios Arampatzakis1, Apostolos Sarris2, Miltiadis Polidorou2, Victor Klinkenberg2, Zeyd Boukhers3, Lingxiao Kong3, Emeri Farinetti4, Fernando Moreno Navarro4, Ioannis Kakogiannos5, Sofia Aparicio6, Javier Ortega6, Fernando Ramonet6, Athos Agapiou7, Stylianos Hadjipetrou7, Kyriakos Michaelides7, Stavros Patsalidis7, Phaedon Kyriakidis7, Apostolos Papakonstantinou7, Demetrios Athanasoulis8, Christos Maris8, Themistoklis Vakoulis9, Tomoki Nagata9, Katrin Beyer9, and Savvas Saloustros9 1Athena Research Center, Greece 2University of Cyprus, Cyprus 3Fraunhofer FIT, Germany 4Roma Tre University, Italy 5Worldsensing, Spain 6ITEFI-CSIC, Spain 7Cyprus University of Technology, Cyprus 8Ephorate of Antiquities of the Cyclades, Greece 9´ Ecole Polytechnique F´ed´erale de Lausanne, Switzerland Abstract The preservation of cultural heritage faces increasing challenges from environmental, climatic, and anthropogenic pressures. The ARGUS Horizon Europe project addresses these challenges by proposing an innovative predictive preservation approach that leverages AI-driven digital twins and multimodal data fusion to assess and mitigate risks to heritage sites. Our research focuses on developing a sustainable, dynamic decision support system (DSS) that integrates multi-scale on-site and remote sensing data, enabling holistic heritage management, while focusing on the 1 preventive rather than the reactive aspects. The primary research question treated in this paper relates to how digital twins incorporating multi-dimensional data and an ontology background enhance the preservation and adaptive management of cultural heritage in a changing environment. 1 Introduction Cultural heritage is increasingly endangered by a convergence of natural, climatic, and anthropogenic threats. Earthquakes, floods, wildfires, freeze-thaw cycles, and coastal erosion regularly affect heritage structures and landscapes, while human-driven pressures such as urbanization, pollution, mass tourism, and neglect accelerate the deterioration of sites and weaken their resilience (Agapiou et al., 2015; Hadjimitsis et al., 2013; Agapiou, 2017). The combined impacts of climate change and societal transformation amplify these risks, threatening both tangible assets—monuments, buildings, archaeological remains—and intangible values such as identity, memory, and cultural continuity. This is particularly acute in remote or exposed sites, where monitoring capacity is limited and local conservation resources are scarce. Traditional preservation strategies often operate reactively, intervening only after visible damage has occurred. Monitoring approaches, when available, are typically fragmented across institutions, data silos, and disciplinary practices. While important progress has been made in documenting heritage through digitization, 3D modeling, and GISbased inventories, these outputs frequently remain static and disconnected from real-time condition monitoring or decision-making processes, particularly in a preventive manner. In order to safeguard cultural heritage under increasingly volatile conditions, there is a pressing need to move from reactive conservation toward preventive and predictive management, grounded in continuous evidence and structured risk assessment (Michalski and Pedersoli, 2016). The notion of the digital twin, originating in Industry 4.0 and cyber-physical systems, offers a compelling paradigm shift. A digital twin is a dynamic, data-rich replica of a physical asset, continuously updated with multimodal data streams and capable of simulating future conditions, supporting diagnostics, and guiding interventions. In a sense, it is a living digital entity. In cultural heritage, this implies moving beyond static digital models or Heritage Building Information Management (HBIM) representations toward living, responsive systems that integrate sensing, semantics, and simulation (Quattrini et al., 2015; Jouan and Hallot, 2019, 2020; Gabellone, 2022; Dang et al., 2023). Such twins can host multimodal datasets—structural, environmental, remote sensing, and intangible— and provide management professionals with predictive scenarios for threats, as well as tools for communication with the public. In parallel, Niccolucci and collaborators have proposed the Heritage Digital Twin Ontology (HDTO) and its extension, the Reactive Heritage Digital Twin Ontology (RHDTO), which formalize semantic representations of assets, sensors, activators, and decision processes for heritage contexts (Niccolucci et al., 2023; Niccolucci and Felicetti, 2024). These works demonstrate how cultural heritage digital twins can evolve from static documentation to reactive systems, linking monitoring data directly to conservation actions. Europe has invested heavily in this direction through research infrastructures and innovation actions. The ARIADNE and ARIADNEplus projects have demonstrated largescale integration of archaeological datasets following CIDOC CRM-based standards (Geser et al., 2023), while E-RIHS is consolidating facilities, labs, and digital services to provide 2 advanced research infrastructure for heritage science across the continent (Striova et al., 2017). The 4CH project has articulated the concept of a European Competence Center for Cultural Heritage, focusing on cloud-based services and standardization (Medici et al., 2022). In parallel, the Time Machine initiative envisions massive digitization and AI-based analysis of European history (Aigner, 2022), and EUreka3D has recently advanced large-scale 3D digitization pipelines aligned with FAIR principles. Together, these initiatives underscore the importance of data integration, interoperability, and shared infrastructures—yet they typically emphasize either documentation, access, or interoperability, without fully embedding environmental monitoring and risk analysis in their digital frameworks. The ARGUS Horizon Europe project builds on this ecosystem but addresses a distinct gap: the lack of integrated, risk-driven digital twins that combine environmental sensing, AI-based threat analysis, and semantic reasoning for preventive preservation. ARGUS focuses explicitly on remote and vulnerable heritage sites, where risks from climate change, environmental processes, and human activities intersect. Its approach integrates multimodal on-site sensing, remote sensing, and legacy data into a geospatial backbone, enriched by a new ontology (titled PANOPTES) and aligned with existing well-established ontologies like CIDOC CRM (Doerr, 2003). The system operationalizes the ICCROM/CCI ABC method for risk assessment (Michalski and Pedersoli, 2016) and adheres to the FAIR principles for scientific data stewardship (Wilkinson et al., 2016). By doing so, ARGUS creates digital twins that not only mirror the present condition of heritage assets but also simulate future scenarios, inform conservation decisions, and promote inclusive engagement through Living Labs and participatory activities. In this paper, we present the conceptual framework of ARGUS’ digital twin. We describe the vision and objectives, the five pilot sites and their risk contexts, the architecture of the data management and digital twin system, and the integration of the ABC methodology into digital workflows. We conclude with a discussion of challenges and outlook, positioning ARGUS as a contribution toward future-proofing cultural heritage through multimodal digital twins. 2 Vision and Objectives The overarching vision of ARGUS is to establish a next-generation monitoring ecosystem for cultural heritage that enables preventive, rather than reactive, preservation. This vision rests on three pillars: (i) scientific integration of multimodal data streams; (ii) technological innovation in sensing, modeling, and visualization; and (iii) social sustainability through participatory and inclusive practices. By addressing these dimensions together, ARGUS aims to transform cultural heritage management from fragmented and reactive interventions to coherent, evidence-based strategies. To realize this vision, ARGUS develops a multi-scale, multi-modal digital twin for heritage assets. This twin integrates heterogeneous sources—in-situ sensors, remote sensing, legacy GIS datasets, and archival records—into a living digital representation that can be dynamically updated, reasoned upon across scales from building detail to landscape context, and used to simulate future risk scenarios. The project also deploys portable, nondestructive sensing systems that are complemented by remote sensing datasets, including Copernicus Earth Observation data, aerial imagery, and UAV-based surveys. These provide an observational ecosystem capable of capturing both localized and regional processes. 3 Central to the project is the implementation of AI-powered threat analysis and risk assessment pipelines. Building upon the ICCROM/CCI ABC methodology (Michalski and Pedersoli, 2016), ARGUS develops machine learning workflows to analyze sensor data, detect anomalous patterns, and prioritize risks. The integration of AI and semantic ontologies allows these analyses to support both predictive simulations and explainable recommendations for management decisions. At the same time, advanced visualization services are designed to meet the needs of different stakeholders, ranging from GIS dashboards that support professional heritage managers to intuitive interfaces that communicate risks and conservation options to wider audiences. Finally, citizen engagement is embedded within the project through participatory formats such as Living Labs and co-creation events, where citizens, students, and local stakeholders contribute data, test applications, and shape narratives. This ensures long-term sustainability by embedding digital heritage management within community practices. Taken together, these objectives articulate a holistic approach that combines technical innovation with social responsibility. ARGUS seeks not only to improve conservation outcomes, but also to foster a sense of shared stewardship, ensuring that cultural heritage remains a living resource that is accessible and meaningful to diverse communities. To ensure that the ARGUS framework is tested under diverse conditions, five pilot sites were selected across Europe, representing different geographies, chronological horizons, typologies of cultural heritage, and risk factors. Together, these pilots provide a robust testbed for validating data acquisition strategies, risk assessment methodologies, and digital twin implementations. The first site, Delos Island in Greece, a UNESCO World Heritage archaeological site, is highly exposed to coastal erosion, salinity, and marine pollution. Rising sea levels and increasing storm intensity pose direct threats to its structural remains, and ARGUS explores how relevant data can be integrated with new sensing strategies to model threat dynamics. In Baltan´as, Spain, a subterranean wine cellar town, vulnerabilities arise from ventilation issues, humidity imbalance, and landslide risks. Its distinctive underground architecture creates microclimatic instability, and ARGUS pilots wireless sensor networks and humidity monitoring strategies to evaluate thresholds for collapse and mold growth. The Monti Lucretili area in Italy represents an upland cultural landscape exposed to erosion, subsidence, and vegetation overgrowth. With its mix of natural and built elements, it offers an ideal case for testing landscape-scale integration of remote sensing and field-based data. Here ARGUS applies geomorphological modeling and vegetation indices derived from satellite imagery to assess conservation needs. The Abbey of Sant’Antonio di Ranverso, also in Italy, is a medieval monastic complex facing water infiltration, biodeterioration, and the pressures of visitor access. At this site, ARGUS examines the potential of structural and environmental sensors to diagnose infiltration pathways and monitor biological growth on stone surfaces. Finally, Schenkenberg Castle in Switzerland, a mountain fortress, is threatened by freeze–thaw cycles, vegetation-induced damage, and long-term structural stability challenges. ARGUS tests vibration sensors and UAV-based thermal imaging here to capture the seasonal effects of freeze–thaw dynamics on stone integrity. As shown on the map in Figure 1, the pilots span distinct climatic zones and geomorphological settings, which is essential for validating generalizable workflows. By covering such a wide spectrum of risks—from coastal erosion to subterranean collapse and alpine freeze-thaw—the pilots ensure that ARGUS methodologies are validated across contexts. They also highlight the importance of tailoring digital twin strategies to the specific vulnerabilities, scales, and data availability of each site. 4 Figure 1: Geographical distribution of the five ARGUS pilot sites across Europe. 3 Materials and Methods 3.1 Ontology-Driven Multimodal Digital Twins The notion of a digital twin represents a major conceptual and technological shift for cultural heritage. Unlike static digital models, a digital twin is a dynamic, data-rich representation of a physical asset that evolves over time, incorporating multimodal data streams and enabling predictive simulation. For heritage management, this means that the digital representation of a monument or site is not limited to geometry or appearance, but actively reflects its condition, vulnerabilities, and risks (Gabellone, 2022; Jouan and Hallot, 2020; Dang et al., 2023). ARGUS adopts a multimodal digital twin architecture tailored to the needs of cultural heritage. The integration of ontologies into the digital twin framework is not merely a technical detail, but a conceptual choice. Digital twins in engineering domains often operate as closed cyber-physical replicas, focusing on synchronization between a physical asset and its digital model. In cultural heritage, however, the problem space is broader: the asset is not only a material object but also a bearer of historical, social, and cultural values; risks are not only physical but also interpretative and managerial; and preservation actions must be accountable and transparent. The digital twin paradigm extends the ontology approach by adding temporality, realtime data ingestion, and simulation. While ontologies traditionally describe classes and relationships, digital twins bring temporality, real-time data ingestion, and simulation into the picture. The ARGUS digital twin thus transforms PANOPTES from a static semantic schema into a dynamic semantic infrastructure, where data are not only curated but also continuously updated, reasoned upon, and acted upon. This mutual reinforcement— using ontologies to make digital twins meaningful, and digital twins to make ontologies actionable—is central to ARGUS. It allows cultural heritage management to move beyond fragmented datasets or isolated 3D models toward integrated, risk-aware, and decisionsupportive systems. Building on this line of reasoning, ARGUS introduces the PANOPTES ontology, to move beyond geospatial integration toward semantic reasoning. PANOPTES extends the CIDOC CRM standard (Doerr, 2003) and integrates ontologies from the sensor and data provenance domains (SOSA/SSN (Janowicz et al., 2019), PROV-O, OWL-Time, 5 Figure 2: The PANOPTES core ontology concept in ARGUS: connecting measurements, risks, and decisions into a semantically integrated digital twin. GeoSPARQL). This provides a semantic backbone for representing assets, measurements, diagnoses, predictions, threats, and policy-driven decisions in a machine-readable and interoperable manner. By capturing not only data but also their meaning, provenance, and temporal context, the ontology allows AI modules to reason about risks and propose actions in a transparent and explainable way. The conceptual rationale of PANOPTES aligns with and extends the current research trajectory. While HDTO and RHDTO (Niccolucci et al., 2023; Niccolucci and Felicetti, 2024) have established important precedents for embedding sensors, activators, and decision logic into semantic frameworks, PANOPTES emphasizes interoperability with established standards (CIDOC CRM, SOSA/SSN, PROV-O, OWL-Time, GeoSPARQL) and integration into multimodal digital twin predictive (and not reactive) workflows. This positions PANOPTES not only as a semantic schema, but as a dynamic infrastructure where ontological reasoning, real-time monitoring, and predictive preservation converge. The rationale and the core ontology concept of PANOPTES is illustrated in Figure 2. The model integrates instruments, protocols, threats, rules, policies, events, measurements, diagnoses, predictions, and decisions, connecting them to assets and cultural documentation. This semantic structure ensures that multimodal data streams can be interpreted in terms of risks, policies, and conservation actions, rather than remaining disconnected technical records. This representation is simplified for illustration purposes and the deep technical specifics are included in the following analysis. Beyond its alignment with CIDOC CRM and related standards, PANOPTES provides a conceptual framework that grounds the digital twin in semantically meaningful structures. At its core, PANOPTES centers on the heritage asset, which may be a movable object, monument, site, or region. Each asset is represented through a sequence of asset states, understood as temporally and spatially situated snapshots of its existence. An asset state embodies the current manifestation of the asset at a specific location and time, and serves as the live instance of the digital twin. Every asset state is associated with cultural documentation, i.e. domain experts’ descriptions spanning history, conservation, provenance, and other disciplinary perspectives. The state is also linked to measurements, obtained through instruments and protocols and reflecting physical quantities such as temperature, humidity, vibration, or spectral reflectance. Measurements may be point values from sensors, multi-dimensional datasets such as thermography, or remote-sensing imagery. These are interpreted by computational models, which infer meaningful events (e.g. a rainfall episode, an earthquake shock, or a humidity excursion) and produce predictions about possible future conditions (e.g. structural tilt, stone erosion, or mold formation). 6 The framework introduces a policy layer to connect data interpretation with preventive action. Policies define thresholds and acceptable conditions (e.g. humidity outside 45-60% for more than 48 hours is a threat to mural stability). By evaluating events and predictions against such policies, the system identifies threats and formulates decisions—recommendations for preventive interventions, recorded in formal decision contexts. Visualizations at each stage (maps, 3D models, heatmaps, dashboards) integrate measurements, events, and predictions into intuitive forms for both experts and the public. In this way, PANOPTES operationalizes the digital twin not only as a technical data hub but as a semantically grounded ecosystem. Assets, states, measurements, models, events, policies, threats, and decisions are explicitly represented, ensuring that ARGUS twins remain transparent, interoperable, and capable of supporting accountable conservation strategies. In simple words, PANOPTES’ rationale can be summarized into the following: To preserve a heritage asset, its temporally and spatially situated state is culturally documented and continuously monitored, through structured measurements derived via instruments and protocols. These measurements are interpreted by computational models to identify events and generate predictions, and intuitive visualizations. Events are evaluated in light of policy constraints to assess threats, and predictions are incorporated into decision contexts to support semantically grounded preventive actions. To complement the conceptual rationale, PANOPTES is implemented as an OWL ontology with a compact set of core classes and object properties aligned to CIDOC CRM, SOSA/SSN, PROV-O, OWL-Time, and GeoSPARQL. Table 1 and Table 2, together with Figure 2, summarize the structure. PANOPTES aligns •Asset/CulturalDocumentation with CIDOC CRM classes; •Measurement/Instrument/Protocol/PhysicalQuantity with SOSA/SSN; •provenance of Measurement and ComputationalModel with PROV-O; •temporal anchors of AssetState,Measurement,Event, and Prediction with OWLTime; •geospatial anchors with GeoSPARQL. To illustrate how the ontology operates in practice, we present a concrete example from the pilot site of Delos, where coastal salinity poses a major conservation risk. In this scenario, PANOPTES structures the full monitoring and decision-support workflow as linked entities, from the asset and its cultural documentation through measurements, instruments, protocols, events, predictions, policies, threats, and preventive decisions. To make the workflow more intuitive, we briefly narrate the example here. At Delos, conductivity sensors detect a spike in salinity on the marble shoreline during a high-wind marine spray event. The system automatically correlates sensor readings with microclimate data and satellite observations, triggering a salinity excursion event. The ABC-based policy thresholds classify this as a crystallization risk, leading to a preventive recommendation (rinsing and windbreak intervention). Table 3 and Table 4 then show how each step is formally represented in PANOPTES. Table 3 summarizes the core entities involved in a salinity monitoring episode, showing how a specific asset state of the Delos shoreline is linked to conductivity measurements, 7 Class Role in PANOPTES Asset Cultural entity under stewardship (movable or immovable: object, monument, site, region). Root of the twin. AssetState Time–space specific manifestation of an Asset; live instance of the twin at a given instant and location (the place to which time/geometry anchors attach). CulturalDocumentation Domain-expert descriptions linked to an Asset or AssetState (history, conservation, provenance), which may vary across states. PhysicalQuantity Measurable property relating to an AssetState (e.g., temperature, RH, electrical conductivity). Measurement Observation/acquisition about an AssetState (point sensor value, image, 3D model, thermography, satellite scene). Instrument Device/system producing a Measurement (e.g., thermistor, humidity sensor, UAV, multispectral camera, satellite). Protocol Procedure/recipe used by an Instrument to obtain a Measurement. ComputationalModel Formal method that interprets measurements to infer Events and produce Predictions; may also drive Visualization. Event Interpreted occurrence at an AssetState (e.g., rainfall episode, humidity excursion, earthquake shock); not necessarily damaging. Prediction Forward-looking estimate of a possible future state or condition (e.g., structural tilt, stone erosion, mold formation). Policy Formalized domain rule expressing acceptable ranges, thresholds, or objectives for evaluation (e.g., RH 45–60% for murals). Threat Semantic interpretation that an Event (evaluated against a Policy) indicates potential risk/degradation. Decision Preventive action recommendation derived by evaluating aPrediction against a Policy (optionally recorded in a DecisionContext). Visualization Spatio-temporal presentation aggregating measurements, events, predictions (maps, 3D, heatmaps, dashboards). PolicyEventLink (optional) Record node linking a specific Event to a Policy with evaluation timestamp and resulting Threat. DecisionContext (optional) Record node linking a Prediction to a Policy with evaluation timestamp and resulting Decision. Table 1: PANOPTES core classes (with two optional provenance record classes). instruments and protocols, cultural documentation, and computational models that infer events and generate predictions. Each element is temporally and spatially anchored, with explicit provenance and standards-based encoding (e.g. ISO-8601 for time, WKT for coordinates). Building on this, Table 4 demonstrates how the ontology records the evaluation chain in which events are assessed against domain-specific policies, producing threats and triggering preventive decisions. In the Delos case, measured electrical conductivity values exceeding thresholds defined in conservation policies lead to the classification of a high salt crystallization risk, which in turn generates a preventive decision recommending rins8 Object Property Semantics (domain →range) hasState Asset →AssetState isStateOf (inverse) AssetState →Asset documentedBy Asset/AssetState →CulturalDocumentation hasMeasurement AssetState →Measurement usesInstrument Measurement →Instrument followsProtocol Measurement →Protocol observedProperty Measurement →PhysicalQuantity interpretedBy Measurement →ComputationalModel infersEvent ComputationalModel →Event producesPrediction ComputationalModel →Prediction evaluatedAgainst Event/Prediction →Policy resultsInThreat Event evaluatedAgainst −−−−−−−−−−−→ Policy →Threat resultsInDecision Prediction evaluatedAgainst −−−−−−−−−−−→ Policy →Decision visualizedAs AssetState/Measurement/Event/Prediction → Visualization hasTime AssetState/Measurement/Event/Prediction → time:Instant hasGeometry AssetState/Measurement →geo:Geometry linksEvent/linksPolicy/linksThreat (optional record) PolicyEventLink →Event/Policy/Threat linksPrediction/linksPolicy/linksDecision (optional record) DecisionContext →Prediction/Policy/Decision Table 2: Key PANOPTES object properties (relationships). ing protocols and protective measures. Provenance links ensure that each decision is transparently grounded in the preceding measurements, models, and policies. These examples highlight how PANOPTES moves beyond abstract modeling by capturing the entire lifecycle of heritage risk management in a machine-readable structure. Assets are not only documented but also continuously observed, interpreted, and acted upon within a semantic framework, ensuring both transparency for experts and interoperability with external infrastructures (see subsection 3.3 for the risk assessment framework that complements this ontology-based representation). The integration of those various layers of documentation transforms the ARGUS digital twin into more than a passive repository. It becomes an active decision-support system that combines multiple functions. Through continuous ingestion of sensor streams the system supports real-time monitoring of asset conditions. Automated workflows then provide AI-supported diagnostics, detecting anomalies and inferring their potential causes—for instance, correlating increases in humidity with rainfall events or patterns of visitor traffic. On this basis, the twin enables the predictive assessment of future risks, including sea-level rise, freeze-thaw cycles, or vegetation overgrowth and structural threats, drawing upon both historical trends and trajectory forecasts. This approach also addresses a long-standing challenge in digital heritage: the disconnection between scientific monitoring data and management decision-making. By embedding the ICCROM/CCI ABC risk model directly into the semantic layer, ARGUS ensures that quantified risks can be linked to both measurements and management actions. Moreover, by adhering to FAIR principles (Wilkinson et al., 2016) and leveraging established ontological standards, the system aligns with European efforts toward inter9 Data, script and code availability No data, scripts, or code were generated or analysed in this study. Supplementary information No supplementary information accompanies this article. Acknowledgments A preprint version of this article has been peer-reviewed and recommended by PCI Archaeo (https://doi.org/10.24072/pci.archaeo.100624). Funding This work has been supported by the ARGUS EU project (Grant Agreement No. 101132308), funded by the European Union. Conflict of interest disclosure The authors declare they have no conflict of interest relating to the content of this article. References Agapiou, A. (2017). Remote sensing heritage in a petabyte-scale: Satellite data and heritage earth engine applications. International Journal of Digital Earth, 10(1):85– 102. Agapiou, A., Lysandrou, V., Alexakis, D. D., Themistocleous, K., Cuca, B., Argyriou, A., Sarris, A., and Hadjimitsis, D. G. (2015). Cultural heritage management and monitoring using remote sensing data and gis: The case study of paphos area, cyprus. Computers, Environment and Urban Systems, 54:230–239. Aigner, T. (2022). Big data of the past comes true – the time machine initiative. In Proceedings of the 27th International Conference on Cultural Heritage and New Technologies (CHNT 2022), Vienna. Propylaeum. Open proceedings paper. Dang, X., Liu, W., Chen, X., Li, S., and Yu, J. (2023). Digital twin applications on cultural world heritage sites in china: A state-of-the-art overview. Journal of Cultural Heritage, 64:1–14. Doerr, M. (2003). The CIDOC CRM—an ontological approach to semantic interoperability of metadata. AI Magazine, 24(3):75–92. Gabellone, F. (2022). Digital twin: a new perspective for cultural heritage management and fruition. Acta IMEKO, 11(1):1–7. 16 Geser, G., Richards, J., Tudhope, D., Vlachidis, A., Jeffrey, S., and Wright, H. (2023). Achievements of the ariadne initiative for archaeological data sharing and research with the ariadneplus project. Internet Archaeology, 64. Hadjimitsis, D. G., Agapiou, A., Alexakis, D. D., and Sarris, A. (2013). Exploring natural and anthropogenic risk for cultural heritage in cyprus using remote sensing and gis. International Journal of Digital Earth, 6(2):115–142. Janowicz, K., Haller, A., Cox, S., Phuoc, D. L., and Taylor, K. (2019). Sosa: A lightweight ontology for sensors, observations, samples, and actuators. Journal of Web Semantics, 56:1–10. Jouan, P. and Hallot, P. (2019). Digital twin: A hbim-based methodology to support preventive conservation of historic assets through heritage significance awareness. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, volume XLII-2/W15, pages 609–615. Jouan, P. and Hallot, P. (2020). Digital twin: Research framework to support preventive conservation policies. ISPRS International Journal of Geo-Information, 9(4):228. Medici, M., Costantini, A., and Niccolucci, F. (2022). D3.1 – design of the CH cloud and 4ch platform. Technical report, 4CH Project. Michalski, S. and Pedersoli, J. L. J. (2016). The abc method: A risk management approach to the preservation of cultural heritage. Technical report, ICCROM and Canadian Conservation Institute, Rome, Ottawa. Open access manual. Niccolucci, F. and Felicetti, A. (2024). Digital twin sensors in cultural heritage ontology applications. Sensors, 24(12):3978. Niccolucci, F., Markhoff, B., Theodoridou, M., Felicetti, A., and Hermon, S. (2023). The heritage digital twin: a bicycle made for two. the integration of digital methodologies into cultural heritage research. Open Res Europe. Quattrini, R., Malinverni, E. S., Clini, P., Nespeca, R., and Orlietti, E. (2015). From tls to hbim. high quality semantically-aware 3d modeling of complex architecture. In The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, volume XL-5/W4, pages 367–374. Striova, J., Rosi, F., Pezzati, L., Cucci, C., et al. (2017). The european research infrastructure for heritage science (e-rihs). In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, volume XLII-2/W5, pages 661–667. Wilkinson, M. D., Dumontier, M., Aalbersberg, I. J. J., Appleton, G., Axton, M., Baak, A., Blomberg, N., and et al. (2016). The fair guiding principles for scientific data management and stewardship. Scientific Data, 3:160018. 17