Skip to content
100% online eventFree registration · Book your placeDEADLINE: Submission of Camera-Ready Papers13 December 2026Details
EuroMed2026

Programme

EuroMed2026 Conference Programme

All sessions will take place online. Registration is free and booking is required via Crowdcast. All times given in Central European Time (CET)

Conference days

Monday 2nd November 2026

All times in CET · 100% online

  1. 07:40

    Admission to the online EuroMed Auditorium

  2. 08:00

    Opening Ceremony Speeches

    • Prof. Nicolas Tsapatsoulis, Vice Rector of Academic Affairs and Research, Cyprus University of Technology
    • Mr. Noel Zammit, CEO, Heritage Malta
  3. 08:15

    Keynote Speaker – David Frank

  4. 08:45
  5. 11:00

    Coffee Break

  6. 11:15
  7. 12:30

    Lunch Break (1 hour)

  8. 13:30
  9. 15:30

    Coffee Break

  10. 15:45

    Session 1: High-Quality Holistic Data Acquisition in Cultural Heritage

    • From 3D Models to Spatial Knowledge: Enriched Archaeological Digitisation in AUTOMATA

      Prof. Gabriele Gattiglia, University of Pisa, Italy

      Abstract

      Why digitise archaeological artefacts in 3D? Beyond visual documentation, three-dimensional models can serve as spatial frameworks that connect, interrogate and reuse heterogeneous forms of archaeological evidence. The Horizon Europe AUTOMATA project explores this potential for ceramic and lithic artefacts, combining AI-augmented robotics, high-resolution photogrammetry and non-invasive archaeometric analysis.

    • Project Meta Paradise: The Digital Twin Transition in Cultural Heritage Through Multimodal 3D Reconstruction and Spatial Mapping of Michelangelo Pistoletto’s Masterpieces

      Prof. Paolo Atzori, University of Udine, Italy · Paper ID 151

      Abstract

      This paper presents "Meta Paradise," an innovative operational framework for the digital and ecological twin transition within contemporary art museums, developed in collaboration with the Pistoletto Foundation. Funded by the European Union under the NextGenerationEU initiative through the Italian Ministry of Culture (PNRR), this project investigates the democratization of contemporary art through advanced extended reality (XR) deployment, transforming the traditional museum into an intelligent communicator within the global Infosphere. The study delineates a comprehensive, multimodal digitizing methodology tailored to the unique geometric, material, and spatial complexities of three seminal works by Michelangelo Pistoletto: Venere degli stracci, Terzo Paradiso, and Metrocubo d’Infinito.

      Our production pipeline integrates multimodal data acquisition strategies including 50-megapixel photogrammetry, high-frame-rate 4K video, and 8K spherical imaging. In the post-production phase, distinct computational methodologies were custom-tailored to the specific nature of each artwork. For the complex, multi-textured surface of Venere degli stracci, a comparative analysis was executed across standard photogrammetry algorithms (Reality Capture and Polycam) and neural rendering techniques, specifically Neural Radiance Fields (NeRF) and Gaussian Splatting, to maximize asset scalability and detail retention. For the reflective and precise geometric structure of Metrocubo d’Infinito, a true 3D polygon mesh was reconstructed using advanced modeling software grounded on calibrated photographic references, enabling direct surface texture-rendering and optimized real-time deployment via web engines like Three.js and Babylon.js. Conversely, the expansive, 80-meter structural layout of Terzo Paradiso was mapped using high-resolution 8K equirectangular stitching for immersive spatial experiences.

      The resulting interactive prototypes successfully bridge the physical-digital divide, offering immersive, device-agnostic WebXR and VR compatibility. The paper concludes by evaluating the systemic efficiency of these diverse 3D and volumetric paradigms, presenting a scalable technological blueprint for international cultural institutions seeking inclusive, globally accessible museum models.

    • GIS for HBIM

      Mr. Tim Camilleri, GeoSys, Malta

      Abstract

      The innovation lies in turning a digital replica into a living tool for cultural heritage management. By connecting 3D models and geographic information with collections records, operational systems and live sensor data, Living Digital Twins allow museums and heritage sites to understand their spaces, assets and changing conditions within a shared environment.

      The proposed demonstration will illustrate how these connections can support practical decisions — for example, identifying environmental conditions that could threaten an artefact, locating the affected collection items and informing a conservation response. This connects digital documentation with day-to-day preservation, while creating opportunities for richer visitor interpretation and more accessible engagement with heritage.

    • Trustworthy Geometry: Metrological Validation and Digital Provenance as Foundations for the 3D Documentation of Historical Keyboard Instruments

      Dmytro Nykonenko, University of Pavia, Italy · Paper ID 226

      Abstract

      Historical keyboard instruments are a demanding case for 3D documentation: a single object can combine structural elements more than two meters long with hundreds of small, interacting mechanical components whose function depends on sub-millimeter geometry. Within the ERC-Synergy project REM@KE (Reconstructing Embodied Musical Knowledge at the Keyboard), 3D scanning provides the geometric foundation of a multimodal documentation protocol that integrates photogrammetry, technical photography, UV/IR imaging, and CT data. If this foundation is to support conservation, organological research, and computational modeling, and to be reused or re-acquired decades from now, two questions must be answered: how accurate is the geometry, and how was it produced?

      In this contribution, we present the 3D component of the REM@KE protocol as an answer to both. First, we describe a low-cost metrological framework for selecting and validating scanners before they are used on heritage objects. A 2.5 m steel reference stand carrying geometric primitives and reference spheres was designed to reproduce the scale of the instruments, with continuous temperature monitoring to keep thermal expansion within tolerance. Six handheld scanners (four laser-based and two structured-light) were tested in five controlled sessions using three-sphere distance measurements, deviation mapping against the nominal geometry, and resolution tests on wooden specimens and historical key levers. The results show that manufacturer specifications are a poor predictor of performance at instrument scale. On this basis, working parameters were defined and subsequently validated on a sixteenth-century virginal used as a training object.

      Second, we argue that a mesh is not only geometry but the result of a process, and that the operator’s decisions on resolution, movement, distance, and real-time assessment are part of that process. Each acquisition is therefore modeled with CIDOC-CRM and its CRMdig and CRMsci extensions as both a digitization event and a scientific measurement, linking the measured component, the device, the software version, the raw point cloud, and the derived mesh. This provenance chain makes it possible, when a component is scanned again after a long interval, to determine whether an observed difference belongs to the instrument itself or to the acquisition pipeline.

      Together, metrological validation and explicit provenance turn 3D models into trustworthy, comparable, and reusable records.

    • Building User-Centred Archaeological Information Systems: Fifteen Years of Experience from Nea Paphos

      Wojciech Ostrowski, Warsaw University of Technology, Poland · Paper ID 221

      Abstract

      Over the last two decades, archaeology has undergone a profound digital transformation driven by the widespread adoption of digital documentation methods and the rapid growth of research data generated during excavations. While technologies such as digital photogrammetry, Reflectance Transformation Imaging (RTI), Geographic Information Systems (GIS), and 3D modelling have significantly improved the quality and quantity of archaeological documentation, they have also created new challenges related to data management, interoperability, accessibility, and long-term preservation. This paper presents fifteen years of experience in developing and implementing digital workflows for archaeological documentation and data management within the Jagiellonian University Archaeological Expedition at Nea Paphos, Cyprus, a UNESCO World Heritage Site.

      The presentation traces the evolution of the project's digital ecosystem from an offline GIS database established in 2011 to a web-based archaeological information system built upon the Arches platform (https://www.archesproject.org/) and deployed since 2020. Particular attention is given to the motivations behind successive technological choices, the practical challenges encountered during implementation, and the lessons learned from managing increasingly complex and heterogeneous datasets generated by long-term archaeological fieldwork.

      The development of the system reflects broader transformations within archaeological practice. For more than a decade, photogrammetry has been routinely used to document excavation contexts, resulting in the accumulation of extensive collections of 3D models, orthophotomosaics and digital surface models (DSMs). To address emerging research and documentation needs, a series of extensions developed between 2024 and 2026 significantly expanded the capabilities of the Arches-based platform. These enhancements introduced support for local coordinate systems, native integration of 3D documentation products, management of orthophotos and DSMs, workflows for importing specialist datasets such as ceramic analyses, and interactive presentation of RTI visualisations associated with individual artefacts and finds.

      Beyond technical development, the paper critically evaluates the human and organisational dimensions of digital archaeology. The experience of Nea Paphos demonstrates that the success of archaeological information systems depends not only on technological sophistication but also on their ability to support diverse user groups. The project team includes students, academic researchers, professional archaeologists, heritage specialists, and experts from multiple disciplines, all of whom require intuitive access to shared information resources. Consequently, the design philosophy gradually evolved towards user-centred solutions that facilitate collaborative work, data sharing, and knowledge exchange within international research environments.

      The paper also discusses the migration of data structures towards CIDOC CRM-based semantic modelling and reflects on the benefits and limitations of ontology-driven approaches in the context of archaeological fieldwork. Finally, the experiences gained through the development of the Nea Paphos digital infrastructure are presented as the foundation for the Mare Nostrum Platform currently being built at the Institute of Archaeology of the Jagiellonian University. The platform combines a centrally controlled thesaurus with multiple Arches instances dedicated to individual archaeological projects and integrates Recogito as a collaborative environment for annotating textual sources, capturing knowledge, and supporting scholarly teamwork. By critically assessing fifteen years of experimentation, adaptation, and innovation, this contribution highlights both the opportunities and challenges involved in creating sustainable, interoperable, and user-oriented digital infrastructures for archaeological research and cultural heritage management.

    • 3D paradata: how far is too far when contextualising a 3D digitisation project?

      Paraskevi Papadopoulou, BnF, France · Paper ID 207

      Abstract

      This article explores the need and limitations of detailed paradata in the context of a 3D digitization project involving a scaled physical model of a slave ship. We describe how, through dialogue among curators, 3D modeling experts, and conservation and digital preservation specialists, we defined the project’s deliverables and the iterative methodology used to produce the model. The article emphasizes the importance of transparency, not only between data providers and curators but also between the institutions disseminating the model and their audiences—whether the general public or academic researchers. It proposes some practical solutions to address this complex challenge and raises the urgent question of establishing common standards for paradata dissemination.

    Questions and Answers

  11. 17:00

    Conclusion end of Day 1 EuroMed 2026

Tuesday 3rd November 2026

All times in CET · 100% online

08:00–09:30

Session 2: Ethical Issue in the documentation of the past

  1. From principles to practice: UNESCO’s ethical approach to digital technologies for cultural heritage protection

    Anne Besançon, UNESCO, France · Paper ID 69

    Abstract

    As the United Nations specialized agency with a global mandate for culture, UNESCO is at the forefront of international efforts to protect and safeguard cultural heritage in all its forms. Guided by a robust normative framework comprising six complementary cultural conventions, as well as recommendations, charters, and other standard-setting instruments, including the 2003 Charter on the Preservation of Digital Heritage and the 2021 Recommendation on the Ethics of Artificial Intelligence, UNESCO promotes an ethical and human-centred approach to the use of technology in heritage protection. Through its New Technologies Initiative under the 1954 Hague Convention, UNESCO is developing innovative tools such as multi-user virtual exploration platforms, AI-driven applications for heritage professionals, and guidance on the protection of cultural property in cyberspace. In partnership with INTERPOL's ID-Art application and other law enforcement partners, UNESCO has also created the Virtual Museum of Stolen Objects, an immersive digital environment that raises public awareness and supports efforts to combat the illicit trafficking of cultural property.

    Building on this experience, the present paper explores both the opportunities and the challenges that emerging technologies present for the protection of cultural property, while emphasizing the importance of maintaining an ethical and inclusive approach. As an international organization serving all Member States, UNESCO must ensure that the tools and solutions it develops are accessible, relevant, and beneficial to a diverse range of stakeholders worldwide. The paper argues that, by adopting a people-centred approach and adhering to the principle of leaving no one behind, new technologies can be effectively harnessed to broaden access to culture, enhance awareness of cultural heritage's significance and vulnerability, and strengthen global capacities for its protection.

    Through applications ranging from early warning systems and stakeholder training to the integration of diverse datasets and advanced analytical tools, the international community is today better equipped than ever to anticipate, prevent, and respond to threats affecting our shared cultural heritage.

  2. Documenting AI use in 3D heritage digitisation and modelling

    Isto Huvila, Uppsala University, Sweden · Paper ID 67

    Abstract

    The integration of artificial intelligence (AI) into 3D heritage digitisation and modelling presents both significant opportunities and complex challenges. While AI enhances efficiency and interpretative potential, it also intensifies longstanding concerns regarding transparency in technological processes. Central to this discourse is the concept of paradata, which addresses the documentation of intellectual and procedural decisions in digital heritage work. This chapter explores the implications of AI for paradata, drawing on emerging literature from both 3D heritage and archival science. It argues that documenting AI use requires a nuanced understanding of decision-making processes of using both 3D and AI technologies. The chapter highlights the limitations of conventional validation strategies in heritage contexts, where definitive benchmarks are often absent. Instead, it advocates for a shift towards putting more emphasis on justifying the plausibility of AI-generated outputs. Cross-pollination between AI and 3D paradata research is proposed as a means to enhance transparency and epistemic accountability in heritage digitisation.

  3. Between Automation and Evidence: The Responsible Use of Artificial Intelligence in Photographic Archives

    Pilar Irala, Universidad San Jorge / Archivo Jalón Ángel, Spain · Paper ID 68

    Abstract

    The integration of artificial intelligence into photographic archives enables the automation of processes such as classification, tagging, object and place recognition, text transcription, description generation, semantic search, and digital restoration. Although these applications may enhance the organisation of and access to large image collections, they also introduce significant risks, including misidentification, algorithmic bias, the loss of contextual information, visual alteration, and the incorporation of unverified inferences into documentary records. These concerns become particularly acute when outputs produced by automated systems are integrated into institutional catalogues and consequently acquire the appearance of authoritative information.

    This research critically examines the tension between automation and evidence in the management of photographic archives. Drawing on an interdisciplinary literature review and a qualitative analysis of the principal forms of algorithmic intervention, it proposes a taxonomy that distinguishes between processes of organisation, extraction, inference, interpretation, transformation, and content generation. On the basis of this taxonomy, the article develops a model for the responsible use of artificial intelligence founded on the preservation of original materials, data provenance, the clear differentiation between verified and inferred information, human oversight proportionate to the level of risk, the reversibility of interventions, and transparency towards users.

    The proposal concludes that the reliability of artificial intelligence in photographic archives depends not only on technical accuracy, but also on the capacity of institutions to document, validate, and communicate algorithmically generated outputs without conflating computational probability with historical evidence.

  4. Negotiating the Ethical Use of AI in Museums: Institutional Logics, Professional Authority, and Distributed Governance

    Elīna Vikmane, Latvian Academy of Culture, Latvia · Paper ID 108

    Abstract

    The relevance of the thesis lies in the conflicting love-hate relationship between the digital sphere and museums (Landi & Marras, 2021). On the one hand, we perceive museums as “conservative, risk-averse field” (Baldwin & Ackerson, 2017, pp. 162–163) or as places “of originals and firsthand or personal experience, [that] do not need AI” (Thiel, 2023, 84-85). On the other hand, they have proven particularly adept at adopting AI (Bareither, 2023, p. 99) to respond to societal need and expactations. Complexities arise from difference in institutional logics - “the research and development in the AI field follows a highly flexible logic different than institutional processes and needs” (Thiel, 2023, 85). This study aims to examine how museums negotiate tensions between established professional ethics and AI cabalities through decisions they make when experimenting with, implementing and governing AI in museum practice?

    Institutional logics in this sense “determine both what should be done in order to achieve desired outcomes and what is good to do” (Haveman et al., 2023, p. 1154). Often, what is considered “good” is translated into codes of conduct and ethical principles guiding responsible action. For many museums, this is long standing international ICOM Code of Ethics as “a set of core principles to guide action in museums” that articulates standards that museum professionals believe in, share, and should uphold” (ICOM, 2026, p. 1). Its latest edition does not specifically address AI ethics. Outside museum sector overwhelming number of more than 160 AI ethics guidelines have been published, leading to repetition and redundancy (Floridi, 2023, p. 58). In additional, high-level principles can be difficult to translate into practice (Seger, 2022, p. 45).

    We seek to find the answer of how museums negotiate tensions through investigating a range of AI practices in museums through 15 museum cases from 11 European countries, representing different geographical, institutional and AI application contexts. The analysis draws on eighteen semi-structured, in-depth interviews with museum professionals, conducted between May and August 2026.

    The findings are synthesised across the following three areas of negotiation: (1) precaution vs experimentation as institutional logics, (2) professional authority vs AI autonomy, and (3) individual responsibility vs distributed AI governance. First, museums often begin experimenting with AI using content that is already public or at least curatorially approved. Hesitation stems primarily from normative uncertainty, which museums address by strengthening staff AI literacy and developing new individual internal policies and manifestos. Second, professional authority is negotiated through varying degrees of human oversight: routine processes and exploratory uses may be delegated to AI, whereas audience-facing AI applications generally receive great human scrutiny. Third, accountability is negotiated through distributed ethical governance. Museums admit encountering ethical challenges arising from AI opportunities that they have not previously faced. AI governance is often extended beyond individual museum employee and an institution across a wider ecosystem in which diverse museum employees, funders, universities, libraries, professional networks, and visitors contribute to shaping ethical AI practices in museums.

    Acknowledgment: The authors gratefully acknowledge the Network of European Museum Organisations (NEMO), which commissioned and funded the NEMO Working Group Digital Transformation report on artificial intelligence in museums, titled "Museum voices on AI in 2026". Data collected for the report form the empirical basis of this paper with written consent from informants and are analysed here within a distinct academic framework.

  5. When Does a Heritage Reconstruction Become a Deep Fake? Paradata as the Disclosure Record under the EU AI Act

    Kaël E. Khoram, Independent Researcher, France · Paper ID 147

    Abstract

    This paper asks when an AI-assisted heritage reconstruction becomes a deep fake under the EU AI Act, and argues that paradata, extended to AI-assisted steps, can connect the scholarly transparency the field already practises with the disclosure the law now requires. Since 2 August 2026, deployers must disclose AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful (AI Act, Articles 3(60) and 50(4)). Generative AI now completes architectural elements, fills in surfaces and animates scenes in reconstructions whose renders and walkthroughs are image and video content, and a museum or heritage agency using such tools is a deployer. A typical case is a museum film that shows a generated façade as the building’s original state.

    Applying the rule to reconstruction raises four difficulties. The field already has a labelling norm: the London Charter (2009) asks that users be told whether a visualisation shows an existing state, an evidence-based restoration or a hypothetical reconstruction, and that the decisions behind it be documented as paradata. Such a label may affect whether viewers mistake a reconstruction for an authentic record, but does not necessarily disclose that AI produced it. The definition refers to “existing” objects, and a destroyed monument may not qualify. An interactive 3D model, unlike its renders, may not count as image or video. And the deployer’s duty has no exception for standard editing, so even routine processing must be assessed against the definition.

    The closest work on each side leaves these questions open: a 2026 review of generative AI in 3D reconstruction (Lombardi et al.) does not discuss the AI Act, and legal analysis of the definition (Meding and Sorge, 2025) works from photography. The paper reads the Act with the Commission’s guidance and tests it against the practices that review describes. It proposes three tools: a classification of reconstruction content as outside or within the definition, with the disclosure each case requires, including artistic works; a minimal paradata record for AI-assisted steps (what changed, why and with what uncertainty; system; sources; human review; presentation context), usable as a quality certificate, that can feed both scholarly documentation and public disclosure; and a proportionality principle matching documentation to the consequences of each use, without reducing the statutory duty. The working hypothesis is that most AI-assisted steps in documentation fall outside the definition, while generated completions presented without context fall within it.

  6. AI driven risks in Art and Antiquities provenance research and due diligence evolution

    Mariusz Wiśniewski, Ministry of Culture and National Heritage, Poland · Paper ID 176

    Abstract

    For decades Poland has worked on the recognition, search, and return of its long-lost treasures of movable cultural heritage, which left the country due to its turbulent past. In the most recent past, the efforts coordinated by the Ministry of Culture and National Heritage have been greatly supplemented and made more efficient by the introduction of image search and image recognition technology and market monitoring workflows based on a combination of expert knowledge with technological solutions. Specifically, the most promising aspect of this work was the evolving use of image crawlers, which allowed for monitoring of lost artifacts while following personal data protection.

    In this paper the author wishes to point to new, emerging threats connected to the advent of popular AI systems and connected AI agents and data crawling efforts of technological startups as well as established ‘tech giants.'

    These are as follows connected to the idea of the internet and its basis of operation. Crawlers, which include image recognition crawlers used by state agencies like the Ministry of Culture, rely on willingly granted access. The aforementioned rights of access are in the age of data mining and unrestricted if not rogue AI agents increasingly revoked by website administrators and data providers, thus creating operational challenges for established technology used lawfully. We are observing first signs of that phenomenon impacting existing operations and wish to sound first calls to build a solution.

    It is argued that the advent of AI might, and in the author's opinion will also bring a new wave to the provenance tampering phenomenon by virtue of scale, relative ease, and the removal of the knowledge and language barriers and traditionally associated burdens of document forgery connected to access and to inner knowledge.

    It is also argued we might soon face challenges as market guardians of malicious compliance and malicious evasion of monitoring tools, both with the use of AI and excused by the supposed use of AI by other parties.

    The author wishes to start to analyze which of these threats are signs of a new, less transparent web and which should be more closely monitored, as something potentially could affect the efficiency of both law enforcement agencies (LEAs) and supporting or coordinating state agencies like the Ministries of Culture. It is still firmly believed that on a level of singular trafficking case the impact of these new challenges might be manageable if not minuscule, but the ease in which they might grow in scale is potentially overwhelming for future operations of LEAs and effort to contain the problem has to be undertaken. An effort that might include higher responsibility placed on the due diligence process, chain of digital data custody and raw data access, and finally zero tolerance for adversarial or extralegal actions in the provenance work and research.

  7. Unlocking Digital Heritage: Strategic Policy Recommendations for Inclusive Digitisation Management

    Marco Rendina, European Fashion Heritage Association, Italy · Paper ID 121

    Abstract

    Digitising cultural heritage does not, by itself, ensure that it becomes accessible, reusable or socially meaningful. Cultural heritage institutions must also navigate fragmented infrastructures, uncertain rights, unequal capacities and ethical challenges intensified by artificial intelligence. Addressing these interconnected barriers requires a shift from digitisation as a discrete technical activity to the integrated governance of preservation, access and reuse.

    This paper presents evidence-informed policy recommendations developed within the Horizon Europe REEVALUATE project, drawing on multidisciplinary research and practical experimentation across three pilots in fashion, sports and audiovisual heritage, and archaeology. The pilots explore contrasting pathways to reuse: transforming historical garments into interactive experiences and contemporary sewing patterns; incorporating museum collections and audio archives into exhibitions and audiovisual productions; and creating an accessible 3D virtual experience of an archaeological site. Together, they provide a practical lens through which to examine how institutional workflows, technological choices and governance arrangements enable or constrain digital heritage reuse.

    The analysis identifies recurring challenges concerning metadata interoperability, rights clearance, dependence on proprietary platforms, meaningful community participation and the sustainability of digital assets beyond project funding. It translates these findings into five interconnected policy priorities: shared and interoperable infrastructures; ethical access and reuse; intellectual property and legal clarity; inclusive participation; and sustainable institutional operations. Proposed measures include domain-sensitive metadata aligned with FAIR and CARE principles, machine-readable rights and provenance information, human oversight of AI-assisted curation, participatory design, continuous capacity building and multi-year operational funding. Particular attention is given to proportionate implementation pathways for smaller and regional institutions, whose resources and capabilities differ substantially from those of larger organisations.

    The paper’s contribution lies in connecting practical heritage experimentation with an integrated policy agenda that treats technological innovation, ethical stewardship and organisational sustainability as mutually dependent. Rather than presenting the pilots as universally representative, it uses their experiences to inform adaptable proposals for European and national policymakers, funding bodies and heritage practitioners. It argues that successful digital transformation should be assessed not simply by the volume of material digitised or online interactions recorded, but by the conditions it creates for meaningful participation, responsible creative reuse and lasting public value.

  8. Modelling Heritage Knowledge for Generative AI

    Lara Corona, Rosas Museum, Italy · Paper ID 65

    Abstract

    The increasing use of generative artificial intelligence (AI) in cultural heritage raises questions not only about the accuracy, bias, and quality of generated outputs, but also about the knowledge structures those outputs depend on. Before generative systems can retrieve, analyse, or use cultural heritage information, it needs to be selected, formalised, and organised into computational representations. Modelling decisions determine which entities and relationships become computationally accessible. Nevertheless, less attention has been paid to how alternative representations of the same heritage evidence may structure the interpretative possibilities available to generative AI. This study investigates how alternative knowledge-representation strategies applied to the same cultural heritage corpus affect both the structure of heritage knowledge and its subsequent generative interpretation.

    The research used the Rosas Museum in Italy as a case study. The study collected qualitative data from 105 participants, including former miners, relatives of former mine workers, museum professionals, community members, and visitors. Responses were analysed thematically and relationally and formalised as subject-predicate-object structures. From the same corpus, two knowledge graphs were designed: a community-informed model, retaining material, human, social, experiential, memorial, spatial, and interpretative relationships; and an object-centred model, restricting representation to material, geological, spatial, functional, and production-oriented relationships.

    Network analysis demonstrated structural differences. The community-informed graph comprised 24 nodes and 31 semantic relations forming one weakly connected component, whereas the object-centred graph comprised 11 nodes and 11 relations distributed across two weakly connected components. Low-frequency relationships could alter graph connectivity, showing that frequency and relational significance are not equivalent. To examine downstream consequences, each graph was independently supplied to the same generative AI task across 20 repeated generations, producing 40 outputs. Object-centred outputs remained predominantly material, technological, and landscape-oriented, whereas community-informed outputs consistently incorporated labour, family and community, memory, solidarity, practical knowledge, gender, and embodied experience.

    The findings demonstrate that knowledge-graph design can establish different computational structures and influence the interpretative possibilities available to generative AI. For digital cultural heritage, knowledge representation needs to be evaluated not only for technical coherence, but also for the cultural relationships it makes computationally accessible. In conclusion, responsible AI begins upstream of generation, with the representational decisions that constitute the interpretative infrastructures of AI-enabled cultural heritage.

  9. From Integrity to Authenticity: Semantic Provenance for Practitioners and Rights in AI-Mediated Intangible Heritage

    Yeeun Lee, Korea National University of Heritage, South Korea · Paper ID 152

    Abstract

    The way we digitally document cultural heritage is changing fast. It’s no longer just about making 3D models or recording performances—we’re now structuring the cultural context itself using knowledge graphs [1]. For example, projects like Memory Twin are taking this further by combining different types of data and information to create richer digital representations [2]. As these digital resources become more layered and complex, generative AI is opening up new possibilities for how we interpret, combine, and reimagine cultural heritage [3].

    But when original records are used to train or transform AI models, there’s a real risk that the cultural roots and contributors' work get disconnected from the digital results. This makes it harder to judge a digital asset's reliability or how it should be used [3, 4]. While technology can help track origins [5], knowing the technical chain isn’t enough—authenticity and the right to use cultural practices must be considered separately. Because intangible cultural heritage is created through practitioners' lived experience and performances, we must consider both the cultural story behind the asset and its technical production history [4].

    In this study, we introduce a way to track and connect the contributions of Hansan Ramie Weaving practitioners to the creation and use of digital assets, especially through their performances. We call this idea Contextual Provenance Integrity (CPI). To do this, we map out the connections—from the practitioners all the way to the original and new digital materials—using standards like CIDOC CRM [6], CRMdig [7], and PROV-O [8]. We also use ODRL to describe usage conditions in a way computers can understand [9], making it clear how cultural credit and legal rights differ. Each relationship in this network is supported by evidence, organized in a special kind of knowledge graph [10]. Tools like SHACL help check that everything fits together [11], and Neo4j lets us explore these connections in detail [1].

    To test this approach, we compare it with other models that keep the context of how information is passed on separate from legal rights. We check how complete the provenance paths are, how accurate the relationships are, and how well contributors can be identified. Through controlled experiments, we found that even if a digital file stays technically intact, its cultural story can still get lost—something our CPI indicators help reveal. This work offers a new way to evaluate cultural provenance beyond technical checks. It gives us a structure for following practitioner contributions throughout the life of a digital asset, and it points toward better ways to manage cultural data in the age of AI.

  10. Bones, Bytes and Meanings: Building a Human (Remains) Digital Repository

    Carlos Eduardo Moreira Moreira, LABOH - CRIA/NOVA FCSH, Portugal · Paper ID 217

    Abstract

    The delicate nature of human remains and the development of 3D digital models thereof raises ethical concerns around the technology and its results. This is further amplified by the contemporaneous surge in connectivity that has led to their almost automatic publication and dissemination, with little regard to image and data protection. As it is recognized that such models can have extensive scientific value, it was decided to assess the ethical implications of the practice before sharing its results. Since ethics are not set in stone but rather in human minds, an ethnographic methodology was chosen as one of the first research avenues.

    However, models must be made accessible to participants in order to do so. This created a practical and ethical dilemma: how can sensitive digital representations of human remains be made available while still being protected and cared for? Four requirements were identified in order to devise a solution: The repository should be (1) easily accessible to authorised users, (2) inaccessible to unauthorised users, (3) sustainable over the long term, including beyond the duration of the overarching project that created the models and its funding, and (4) not dependent on for profit technology companies whose business models and practices may conflict with underlying ethical principles. A survey of existing repository and hosting options showed that no available solution fulfilled all requirements simultaneously. Therefore, the project adopted an autonomous approach and developed a bespoke repository.

    This communication presents the creative process leading up to it from a human and ethical perspective, focusing on what shaped the creation of the repository and the experience of creating it rather than its technical architecture. It examines how ethical considerations surrounding 3D digital models of human remains were translated into practical requirements for storage, control, access and long-term sustainability. This was intended not only to facilitate consultation with the public, but also to have a long term, in-house solution that may contribute to a broader discussion concerning responsible stewardship and access to 3D digital representations of human remains.

  11. 09:30–09:40

    Coffee Break

  12. 09:40–12:00

    Session 3: Advanced 3D digitisation

  13. Pollocks of the Past: Blending Archival Photos and Art with GenAI+3DGS

    Fadilah Husna Arief, Adelaide University, Australia · Paper ID 53

    Abstract

    We present a workflow that combines Generative AI (GenAI) and 3D Gaussian Splatting (3DGS) to reconstruct virtual heritage experiences from sparse historical photographs and artworks. Rather than pursuing definitive or photorealistic reconstructions, our workflow offers an accessible approach that enables Galleries, Libraries, Archives, and Museums (GLAMs) to transform limited archival collections into interactive, device-agnostic experiences. Using the nearly-demolished Destitute Asylum in Adelaide, Australia, as a case study, our prototype illustrates how the visual characteristics and rendering attributes of 3DGS can expose uncertainty and interpretation, opening opportunities for museums to foster critical reflection and dialogue about the reconstruction of the past rather than presenting a single representation.

  14. Exploring design principles for crowdsourcing 3D Cultural Heritage metadata

    Sofia D. Piven, Adelaide University, Australia · Paper ID 196

    Abstract

    3D Cultural Heritage objects (3DCH) have proliferated across the digital humanities as rapid and user-friendly 3D digitization techniques continue to be adopted by cultural heritage organizations in service of research, education, and promotion goals. However, much published 3DCH lacks adequate metadata, or “data about data”, which describes, explains and otherwise contextualizes 3DCH for users to find and make decisions about how best to use it. Part of the reason is that metadata schemas typically used for cultural heritage collections fall short of adequately describing 3DCH. Without good quality metadata, 3DCH risks becoming uninterpretable and unlinked from the wider CH discourse.

    So far, much attention has been directed to developing extensions and new schemas for describing 3DCH, and visual editors and AI tools to help simplify and scale metadata description and enrichment for authors of 3DCH. However, as the techniques for producing 3DCH and its range of applications diversifies, the task of creating complete, relevant and user-centered metadata becomes increasingly complex. It is costly, if not infeasible, for authors to exhaustively anticipate possible future use cases for 3DCH and describe and continuously enrich 3DCH accordingly. Moreover, in cases where the descriptive categories to express specialized knowledge for 3DCH simply do not exist, schemas take time and collaboration to extend or develop. With the expansion of 3DCH, “top-down” metadata creation on its own becomes increasingly unsustainable.

    By comparison, approaching metadata as a shared responsibility between authors and users of 3DCH suggests new potential solutions. The idea of participatory or “crowdsourced” metadata in cultural heritage is not new, with projects leveraging platforms like Zooniverse for the metadata description of cultural heritage collections. Crowd participation can be structured in different ways to support the execution of a goal, with users contributing to different degrees of depth and complexity. However, how different forms of participation could be put in service of metadata creation is underexamined, let alone in the context of 3DCH collections.

    To fill this gap, based on a tightly scoped review of cultural heritage crowdsourcing typologies and recent cases of cultural heritage metadata crowdsourcing, we propose a framework for designing crowdsourcing for 3DCH metadata. We discuss the potential place of different forms of participation in relation to metadata creation, from supporting existing standards-based metadata description to co-creation of descriptive categories for 3DCH. Finally, we discuss the logistical and infrastructural challenges facing cultural heritage organizations interested in crowdsourcing 3DCH metadata.

    The significance of our framework is that it supports cultural heritage organizations in assessing the suitability of crowdsourcing for 3DCH metadata creation to help offset bottlenecks due to growing volumes of 3DCH.

  15. From Digitisation to Data Sharing: the case of the IMPULSE Project

    Theodora Rontzova, KU Leuven, Belgium · Paper ID 86

    Abstract

    The digitization process does not end with the high quality capture of data. The IMPULSE project (IMMmersive digitization: uPcylcing cULtural heritage towards new reviving StratEgies, funded by Horizon Europe) investigates the reuse of already digitized collections within multi-user virtual environments, bringing together two-dimensional, three-dimensional and time-based digitized assets. In my presentation, I aim to focus on the work carried out within WP3 (STAND), which approaches standardization from the practice-based perspective of heritage institution, taking into account their workflows, resources and diverse levels of expertise. A scoping review of literature, standards, guidelines, projects and platforms centered on the digitization of cultural heritage showed that, while the acquisition of 2D, 3D and 4D data are well supported by existing standards, multi-user virtual environments lack established frameworks and institutional workflows remain largely undocumented in publicly available sources. Building on these findings, guidelines for metadata simplification propose a minimal set of Dublin Core elements combined with a persistent identifier that links each asset to the full record held by its source institution. I will also present our work on paradata- understood as the record of professional decisions and technical conditions behind digitization. Drawing on case studies of 2D and multimodal imaging at KU Leuven Libraries, 3D digitization at Heritage Malta and process modelling with BPMN at the Jagiellonian University, a methodology for paradata creation was developed, structured in three tiers (Minimum Viable, Scientific and Memory Twin) and five phases, from planning and complexity assessment, through metadata strategy, data acquisition and processing, to ingest and publication. At its highest tier, the framework aims at a Memory Twin of the object: a comprehensive paradata record that documents the material and geometric challenges, the environmental conditions and calibration of capture, and every manual or AI-based intervention, so that future researchers can understand each decision and constraint that shaped the digital form of the asset. The tiered structure offers smaller institutions an achievable entry point, while providing a path towards fuller documentation of calibration, quality assurance, uncertainty and manual or AI-based interventions. Currently, these results are being brought together in a Digitization Competence Audit Toolkit, which is intended to help small to medium scale heritage institutions enhance their workflows towards data sharing in novel virtual environments and emerging common European data spaces. By presenting these findings, together with insights from the IMPULSE Final Hackathon held in Athens in June 2026, my presentation will demonstrate how holistic documentation of the digitization process can be made feasible for heritage institutions of all sizes, and how it supports the transparent and sustainable reuse of cultural heritage data in immersive environments.

  16. Interpreting Landscape Memory in the Atlantic Forest: Connecting Archaeological Remains, Vegetation and Historical Land Use in Ilhabela, Brazil

    Barbara van Sebroeck Martins, INPE, Brazil · Paper ID 163

    Abstract

    Tropical forests conceal material traces of historical occupation, challenging archaeological documentation and landscape interpretation. Understanding these landscapes requires moving beyond detecting physical remains to connect spatial evidence with historical land use and environmental transformations. This ongoing study investigates how multi-sensor remote sensing, archaeological inventories and historical contextual information can support the interpretation of landscape memory in Ilhabela, southeastern Brazil.

    The research examines a forest-covered landscape shaped by former sugarcane, coffee and banana cultivation. The methodological framework integrates airborne and UAV-based LiDAR, multispectral imagery, documented archaeological locations and historical information. LiDAR-derived digital terrain and canopy height models, terrain visualisations and multispectral data are examined to investigate archaeological features and their landscape context. Geomorphometric and vegetation-related variables support the characterisation of known archaeological locations and comparison areas.

    Initial investigations identified candidate locations within Ilhabela State Park, three of which were subsequently confirmed through field verification as containing archaeological evidence. These findings illustrate the contribution of integrating remote sensing and existing archaeological knowledge to targeted prospection. Further analysis investigates relationships among archaeological features, terrain, hydrography, historical agricultural uses and contemporary forest structure, distinguishing observed patterns from historical interpretations. Forest structure in areas undergoing secondary succession, together with the occurrence of jackfruit trees and bamboo, is examined as potential evidence of past human land use within the protected landscape.

    Rather than presenting a completed MemoryTwin, the research develops a methodological foundation for connecting digital landscape representations with historical knowledge. By considering archaeological remains alongside vegetation patterns and historical context, this approach supports the interpretation of processes embedded in cultural landscapes and the progressive reuse and enrichment of heritage information across archaeological research, landscape conservation and cultural heritage management.

  17. THE RELIABILITY OF DIGITAL EVIDENCE ON LOST CULTURAL HERITAGE: DATA ORIGIN, RECONSTRUCTION AND UNCERTAINTY IN THE CONTEXT OF THE WAR IN UKRAINE

    Yuliia Trach, Kyiv National University of Culture and Arts, Ukraine · Paper ID 218

    Abstract

    The destruction of cultural heritage due to armed conflicts underscores the importance of evaluating the reliability of digital evidence related to sites that have been damaged, destroyed, or are no longer accessible for direct study. While a 3D model can preserve an object's geometric and visual characteristics, it does not provide insight into the circumstances of its creation, its historical context, or the reliability of the information available about it. In the absence of a physical original, it is crucial to trace the provenance of digital data. It is important to differentiate among recorded characteristics, documentary evidence, and reconstructions, and to verify claims about the object's previous condition. In this study, digital evidence is viewed not just as a three-dimensional model, but as a combination of the digital representation and the available information concerning its creation, sources, and documentary basis. This definition includes both the digital model itself and the data needed to verify its provenance and validity.

    The study aims to develop an analytical approach for assessing the authenticity of digital representations of cultural heritage in the context of military threats and loss of the original, based on Ukrainian practices in 3D documentation. The theoretical framework is drawn from research on digital cultural heritage, data provenance, and the documentation of cultural memory. The concept of the 'Memory Twin' is an important framework for understanding the transition from recording an object's physical characteristics to preserving knowledge of its history and cultural significance. At the same time, the study focuses on a more specific issue: by what means does digital documentation ensure the traceability of evidence and enable the distinction between documented facts and subsequent reconstructions and interpretations. It is essential to distinguish between the visual persuasiveness of a digital representation and the evidential validity of the information it contains. Therefore, even a detailed model requires verification of the origins of the depicted characteristics, particularly if some have been reconstructed from indirect sources.

    The methodology involves a comparative qualitative analysis of two Ukrainian practices in the digital documentation of cultural heritage: the #SaveUkrainianHeritage initiative by the Skeiron team and the digital documentation project associated with the Museum of Ukrainian Victory / Pixelated Realities. The empirical basis consists of available 3D models, project descriptions, metadata and supporting materials, which can be used to establish the origin of the data and the circumstances surrounding the creation of the digital representations. The analysis is carried out according to four criteria: (1) data origin – authorship, time and circumstances of recording; (2) documentary traceability – the availability of information on data collection and processing methods; (3) differentiation of knowledge statuses – the ability to distinguish between recorded characteristics, historical evidence, reconstructions and interpretations; (4) temporal specificity – the ability to establish which state of the object the model reflects and whether its subsequent changes or loss have been documented. A structured comparison makes it possible to determine which characteristics of digital evidence are corroborated by the available documentation, and where uncertainty remains. The comparison does not aim to establish a universal ranking of projects, but focuses on identifying the strengths and limitations of the available documentation for each criterion. This allows uncertainty to be viewed not as an automatic indication of the unreliability of a digital representation, but as a characteristic of the limits of its verifiability. The absence of information in publicly available materials is regarded as a limitation on verification, rather than as evidence of the absence of relevant practices within the project.

    The result is an analytical framework for assessing the authenticity of digital representations of cultural heritage where the ability to verify them against the physical original is limited. The academic contribution lies in operationalising the issue of the traceability of digital evidence and distinguishing between the documented characteristics of an object, reconstructed information and interpretations. The frame allows for an assessment of the extent to which available documentation supports claims regarding the characteristics and history of a lost object. Its application can be broadened to other contexts of armed conflict, in which digital documentation becomes an important source of knowledge about damaged or destroyed cultural heritage.

    Keywords: digital cultural heritage, 3D documentation, data provenance, digital evidence reliability, war-related damage to cultural heritage.

  18. Landscapes Remember: Reconstructing Early Holocene Alaska Through Immersive Technology

    Brooke Hoover, University of Alaska Anchorage, USA · Paper ID 187

    Abstract

    Archaeological landscapes are continuously transformed by environmental processes, often obscuring the spatial relationships that influenced past human occupation, mobility, and resource accessibility. Traditional archaeological mapping and visualization methods provide valuable spatial information but can limit opportunities to examine past landscapes as integrated, immersive environments. This research explores how geographic information systems (GIS), three-dimensional modeling, and virtual reality (VR) can be integrated to reconstruct archaeological landscapes and expand their interpretive potential.

    Focusing on the Tangle Lakes region of Interior Alaska approximately 10,000 years ago, this thesis develops a digitally reconstructed paleolandscape using archaeological research, digital elevation models (DEMs), and environmental data. Through a workflow integrating ArcGIS Pro, TerreSculptor, and Unreal Engine, geographic datasets are transformed into an immersive three-dimensional environment that contextualizes archaeological site locations within reconstructed terrain and hydrological conditions.

    Beyond visualization, this project investigates how immersive landscape reconstruction can support archaeological interpretation by facilitating the examination of site distribution, landscape visibility, potential movement corridors, and human-environment relationships. It also addresses methodological challenges associated with terrain accuracy, environmental reconstruction, and the representation of uncertainty when translating archaeological evidence into virtual environments.

    The significance of this research extends beyond the reconstruction of a single archaeological landscape. By integrating geospatial analysis with immersive technologies, the project demonstrates a methodological approach for making complex archaeological information more accessible, supporting spatial inquiry, and communicating environmental change across deep time. This framework offers potential applications for archaeological research, heritage preservation, education, and the interpretation of culturally significant landscapes where physical access may be limited or environmentally sensitive.

  19. Digitizing And Integrating Above-Ground and Underground Structures Using Low-Cost Photogrammetry: A Byzantine Cistern Case Study

    Mert Okay, Kadir Has University, Turkey · Paper ID 177

    Abstract

    The Historic Areas of Istanbul were inscribed on the World Heritage List in 1985 (ref. 356) [1]. Recent scholarship attributes at least 211 cisterns to the Byzantine city [2], yet the reference literature for these structures remains typological and topographic [3, 4] rather than metric and three-dimensional. Where heritage risk has been systematically assessed in Istanbul, most notably in the cultural heritage component of Istanbul Seismic Risk Mitigation and Emergency Preparedness Project, which evaluated 170 historical buildings [5], underground structures remain metrically undocumented and effectively outside the scope of risk assessment, despite their structural entanglement with the protected fabric above them. No comprehensive study has documented an underground structure in Istanbul integrated with the structure it supports as a single, integrated record. This paper addresses this gap by demonstrating that photogrammetry method can produce a metrically acceptable, integrated 3D record of an underground structure and the above-ground structure it supports, even under the typical constrained conditions of Istanbul cisterns (active commercial use, limited access, and poor lighting). The workflow is tested on the Nakilbent Cistern, constructed in the 5th century and selected because it carries a multi-use above-ground structure and remains accessible to visitors [6, 7]. The point cloud is then translated into an HBIM model at LOD100 and LOD300, in which each structural element is modelled from its own measured geometry and labelled as measured, typologically transferred or interpreted, while its level of detail and grades of generation and accuracy are recorded following the GOA-LOG specifications [8]. The same framework extends to the above-ground structure, and the deviation of the model from the point cloud is evaluated as a prerequisite for its reliable use in structural analysis [9]. The workflow relies on low-cost equipment, tolerates constrained environments and is replicable across comparable cisterns. Extended systematically, such merged records could form an integrated above ground/underground documentation base for the Historic Peninsula, serving as an information source for disaster management and as a pathway through which underground heritage might be incorporated into disaster management plans of the area.

  20. Multiscale 3D Heritage Documentation in Algeria: Insights from the CRAT’s Experience

    Meriem Seghiri, Centre National de Recherche en Aménagement du Territoire (CRAT), Algeria · Paper ID 87

    Abstract

    The diversity of Algeria’s historical and archaeological heritage presents a methodological challenge for digital documentation. How can 3D digitisation techniques be adapted to heritage assets as varied as a funerary monument, a hydraulic structure, a medina (urban heritage), an ancient city, a frontier outpost along the Roman limes, a collection of mosaics, or modern architectural heritage? This paper addresses this question through the experience of the National Centre for Research in Spatial Planning (CRAT), which has digitised more than 40 sites and monuments across Algeria.

    The corpus encompasses funerary monuments, including the Medracen and the Tomb of El Soumaa, a Roman aqueduct, Roman cities such as Tiddis, frontier outposts along the limes, including Tibaaliyane and Amdoukhel, and Ottoman palaces such as the Bey’s Palace. It also includes the urban heritage of Old Mila, collections of mosaics, and the architectural heritage of Constantine 1 University, designed by Oscar Niemeyer. This diversity provides a basis for examining how heritage typology, spatial extent, and accessibility influence data acquisition strategies. It also enables analysis of the complementary roles of terrestrial 3D scanning and UAV surveying, from architectural detail to the relationship between heritage complexes and their surrounding landscape.

    The analysis follows three main strands: planning data acquisition according to the configuration of the remains, integrating terrestrial and aerial observations, and establishing the conditions for data consistency and reuse. The two techniques are compared in terms of their ability to capture complex geometries, areas that are difficult to access, and extensive spatial configurations. Surface coverage, occluded areas, and transitions between scales are central to the discussion.

    The paper then presents a comparative analysis of the documentation requirements specific to each heritage category. Funerary monuments require particular attention to volumetric form and architectural detail. Hydraulic structures call for documenting the continuity of their alignment and their relationship with the terrain. Ancient cities and sites along the limes require linking the recording of remains to an understanding of their territorial organisation. The urban heritage of Old Mila and the mosaic collections require documentation tailored to their scale and material characteristics. The modern heritage of Mentouri 1 University, designed by Oscar Niemeyer, calls for an approach responsive to the specific characteristics of twentieth-century architecture.

    Together, these acquisitions aim to establish a heritage database designed to serve different user profiles and their specific needs in research, conservation, restoration, heritage interpretation and public engagement, and education. This database provides a foundation for digital twin and Heritage Building Information Modelling (HBIM) applications, supporting dynamic management, simulation, and the ongoing documentation of sites and monuments.

    The contribution places an extensive Algerian digitisation experience in a broader methodological context to support coherent digital documentation for research, conservation, and heritage interpretation and public engagement. It situates 3D digitisation within a discussion of the relationships between monument, site, and territory, bringing a North African perspective to the debate on Mediterranean digital heritage.

  21. A Volumetric Video Pipeline for Digital Twins of Traditional Performing Arts: Gaussian Splatting Refinement and Playback of Chunaengmu

    Wonseok Lee, Korea National University of Heritage, South Korea · Paper ID 186

    Abstract

    Traditional performing arts involve movements, costumes, and ornaments that come together and change over time. Photographs and written records have limitations in capturing these changes, while conventional videos make it difficult to examine three-dimensional forms and details outside the recorded viewpoints. Volumetric video allows performances to be viewed from different angles and may serve as a visual foundation for Memory Twins that connect digital records with historical context and knowledge of cultural transmission. However, free-viewpoint observation does not guarantee that the original forms and details are faithfully reproduced. Fine ornaments may appear blurred, and attempts to correct them can reduce image quality in other regions. This study presents a volumetric video processing pipeline for Chunaengmu that uses selective Gaussian Splatting refinement to improve visual fidelity. The pipeline also includes verification of non-target Gaussian attributes, quantitative evaluation, and free-viewpoint playback.

    The pipeline has five stages: input data inspection, verification of frame and camera correspondence, selective Gaussian refinement, evaluation of a partial sequence, and fixed-view and free-viewpoint playback. We used multi-view recordings of Chunaengmu, camera calibration data, and frame-wise PLY files containing Gaussian attributes. The original renderings served as a baseline, and results before and after refinement were compared with the recorded images at matching frames and camera viewpoints.

    The first refinement test improved details of the ceremonial headdress (hwagwan), but image quality in the full-body region decreased. We therefore limited the refinement to Gaussians representing the upper part of the headdress. The selected region was projected onto multiple camera views, and its boundaries were adjusted to reduce the inclusion of the forehead and hair. Some large Gaussians within this region were split, and their positions, rotations, scales, opacities, and base colors were optimized. Gaussians outside the selected region were restored to their original values at each optimization step. Their attributes were checked again after processing.

    We evaluated the method using a sequence of 30 frames. Compared with the original renderings, the average peak signal-to-noise ratio (PSNR) increased by 1.455 dB in the head region and 0.170 dB in the full-body region. Non-target Gaussian attributes remained unchanged in all 30 frames. We also examined continuous comparison videos for changes in the headdress and flickering between frames. These tests assessed image similarity and attribute preservation, but they did not fully evaluate temporal consistency or the overall fidelity of the cultural heritage representation. The refined results were prepared as comparison videos and Gaussian animations for free-viewpoint playback, and processing was extended to the full sequence.

    The main contribution is a refinement procedure that limits changes to selected Gaussians and evaluates both image quality and the preservation of non-target attributes. In the Chunaengmu case, local refinement improved image quality in the head region without changing Gaussian attributes outside the selected area. The workflow connects this refinement process with quantitative evaluation and free-viewpoint playback. It may be useful for improving fine details in volumetric records of traditional performing arts and for future Memory Twin applications that link these records with historical context and transmission knowledge.

  22. Documenting Historic Buildings in UAE Using Gaussian Splatting for Heritage Conservation

    Bilal Dridi, University of Derby, UK · Paper ID 66

    Abstract

    Documenting historic buildings provides an essential foundation for understanding their condition, guiding conservation, and preserving architectural knowledge. However, producing detailed digital records can require substantial time and specialist resources. This paper explores the potential of three-dimensional Gaussian splatting to support efficient heritage documentation in Sharjah, United Arab Emirates, with particular attention to data acquisition and the usefulness of the resulting records for conservation.

    The research draws on the capture of three historic buildings using XGRIDS PortalCam. These buildings serve as case studies for investigating how acquisition decisions can influence documentation quality and recording effort. Building on this practical experience, the paper proposes an evaluation framework addressing capture routes, viewpoint diversity, lighting conditions, occlusion, and targeted recording of significant architectural details. The framework distinguishes visual quality and completeness from geometric accuracy and connects each assessment to a defined conservation task.

    Initial practitioner observations suggest that the workflow offers potential time savings, although their extent requires systematic comparison with an equivalent documentation baseline. Independent validation is also necessary before the resulting models can support dimensional measurement or detailed condition assessment.

    The intended contribution is a practical approach to planning and evaluating Gaussian splatting acquisition for historic buildings. By connecting recording efficiency with evidence quality, the research examines how emerging digital technologies can support heritage conservation while retaining professional judgement, transparency, and accountability.

  23. Applying Gaussian Splatting (3DGS) at a Coastal Maya Site El Meco

    Ashuni Emmanuel Romero Butrón, Instituto Nacional de Antropología e Historia, Mexico · Paper ID 181

    Abstract

    Between 2023 and 2024, the Research and Conservation Project El Meco (PICME) digitally documented building maintenance work and numerous systematic excavations at the El Meco archaeological site in Quintana Roo, Mexico. Our primary goal was to produce three-dimensional digital models that support later analysis and, at the same time, serve as effective tools for disseminating and communicating our work and findings. Using Gaussian Splatting, a volumetric rendering technique, we have engaged both specialized and general audiences with the cultural heritage of the ancient settlement of El Meco in an accessible and visually compelling way.

    Gaussian Splatting builds on decades of research in direct rendering, from Lee Alan Westover’s pioneering work in 1991 to later studies (Botsch et al., 2005), and it was Kerbl et al. (2023) who marked a turning point, strongly influencing the 3D modeling industry and applications in augmented (AR) and virtual reality (VR). The technique represents a scene as splats generated from two-dimensional images, each with spatial coordinates (X, Y, Z), opacity, color, shape, and orientation. These attributes let the resulting models reproduce changes in luminance as the viewing angle shifts, yielding notably photorealistic representations. The technique also processes large volumes of data quickly and produces lightweight models that are compatible with many programs and platforms while preserving realistic real-time rendering.

    We processed several image sets from diverse contexts with Luma Lab, Polycam, Scaniverse, NerfStudio, and Postshot, obtaining comparable results in quality and file size. The clearest differences arose from the capture devices used and the scale of the documented areas. Platforms such as Gracia and Arrival Space allowed us to share the results more simply and realistically, showing El Meco’s heritage to colleagues and the public through VR headsets or online access.

    To date, only a few cases have systematically applied this technique to the documentation and visualization of cultural heritage (Jamil and Brennan, 2025). Implementing it in our project is therefore both a challenge and an opportunity to digitally preserve part of the cultural heritage of El Meco.

  24. Assessing the quality of digital preservation of gold cultural heritage artefacts via use of 3D Gaussian splatting

    Petros Stavroulakis, 3Daedalus, Greece · Paper ID 51

    Abstract

    Gold and gilded artefacts remain among the most difficult subjects in heritage 3D documentation. Their strongly specular, view-dependent reflectance violates the Lambertian assumptions underlying most triangulation-based systems, producing incomplete point clouds, spurious geometry, and a reliance on matting agents that conservation practice does not permit. This paper evaluates 3D Gaussian Splatting (3DGS) as a non-contact alternative and situates its performance against two established active techniques.

    We reconstructed a collection of two types of gold and gilded objects which were copies of museum artefacts, one a crown leaf with very small features and a standard gold coin which was thicker and with features of larger size. Their size ranges were between 20–250 mm in diameter. Each object was recorded three times: with 3DGS from 100 images acquired using Sony α7ΙΙΙ under diffuse lighting; with a Shining 3D Rigil triangulation scanner at 20um; and with an infrared structured-light system (Revopoint Pop2 Scanner). Geometric agreement was quantified through cloud-to-mesh deviation against the laser scanned object which was the one with the most dense point cloud and accurate measurement specs, reporting RMSE, mean signed deviation, and surface coverage. Appearance fidelity was assessed on held-out views using PSNR, SSIM, and LPIPS. Acquisition and processing time, operator burden, and equipment cost were recorded throughout.

    Out of the three 3D reconstruction modalities, 3DGS achieves markedly superior appearance fidelity and near-complete visual coverage which is unaffected by strong reflections and the non Lambertian surface, albeit using a very small amount of actual geometric points. The 3DGS approach therefore is a very promising route for reconstructing gold and gilded for public facing cultural heritage applications such as in museums and exhibitions.

  25. Advanced 3D Digitisation and HBIM-Based Documentation of Hadrian’s Arch in Jerash: Integrating Terrestrial Laser Scanning, Photogrammetry, and UAV Surveys for Structural Analysis

    Alaa Ababnjeh, Autonomous de Barcelona University, Spain · Paper ID 111

    Abstract

    This paper presents a comprehensive methodological framework for the advanced 3D digitization and digital preservation of Hadrian’s Arch in Jerash, Jordan, a monumental Roman structure facing complex structural degradation. By synergizing Terrestrial Laser Scanning (TLS), close-range photogrammetry, and Unmanned Aerial Vehicle (UAV)-based surveying, the study achieves a high-precision, multi-resolution spatial dataset that captures intricate architectural geometries and surface conditions. The acquired dense point clouds and high-resolution orthophotos serve as the geometric foundation for developing a detailed Heritage Building Information Model (HBIM) aligned with international openBIM standards (IFC/ISO 19650) and ICOMOS deterioration terminology. The geometric data is integrated with Finite Element Analysis (FEA) to assess structural vulnerabilities, deformation patterns, and stone decay mechanisms. This integrated approach bridges geomatics engineering and structural conservation, delivering a scalable digital archive that supports preventive conservation, restoration planning, and public engagement for vulnerable archaeological heritage in the Middle East.

  26. A Data-Driven 3D Reconstruction and Geometric Analysis Framework for Morphological Compatibility Assessment of Archaeological Artifacts

    Davide Tosi, Università degli Studi dell'Insubria, Italy · Paper ID 71

    Abstract

    The authentication of archaeological artifacts traditionally combines expert visual inspection, provenance analysis, and physicochemical examination. Three-dimensional morphology can provide complementary, non-destructive evidence, but reproducible workflows connecting photogrammetric reconstruction with quantitative comparison remain insufficiently explored. This paper presents an end-to-end framework for morphological compatibility assessment based on controlled photographic acquisition, 3D reconstruction, geometric descriptor extraction, and point-cloud registration against curated reference specimens. The workflow comprises four intelligent agents, each specialized in one macro-phase: reference dataset construction and baseline curation; multi-view acquisition and preprocessing; photogrammetric reconstruction with feature extraction; and registration-based estimation of morphological coherence. A feasibility study was conducted on a physical Greek Lekythos candidate, using a reference dataset of 69 images of authenticated 5th-century BCE Lekythoi from the Metropolitan Museum of Art, the Cleveland Museum of Art, and the British Museum, with five British Museum vessels providing the morphometric comparison sample, complemented by an auxiliary set of 65 Lekythoi from other chronological periods. From 83 high-resolution photographs, a photogrammetric pipeline produced a textured mesh converted into a dense point cloud of 1,000,000 points. The reconstructed artifact exhibited a height-to-width (slenderness) ratio of 2.50, within the archaeological range of 2.45–2.65 reported for authenticated Classical-period Lekythoi and matching the reference-sample mean. At the same time, Gaussian curvature analysis confirmed high surface regularity with no abrupt geometric discontinuities. Rigid point-to-point ICP registration against an authenticated 3D reference model, performed on uniformly resampled clouds of 50,000 points with an inlier threshold of 0.05 model units, produced a fitness score of 0.9688 (96.88% of target points matched) and an RMSE of 0.0207 model units, i.e., a normalized error of about 3.02% of the vessel height. Under the adopted registration configuration, the target model exhibited close geometric correspondence with the selected reference model. Given the single-target artifact and the limited reference set, the results indicate technical feasibility rather than independent proof of authenticity. The framework is intended as a quantitative, reproducible, and non-destructive decision-support component complementing archaeological expertise, provenance research, and archaeometric analysis.

  27. From 3D Documentation to Historical Reinterpretation: Re-reading the Bas-Reliefs of Banteay Chhmar in the Angkorian Kingdom

    Ichita Shimoda, University of Tsukuba, Japan · Paper ID 115

    Abstract

    From the late twelfth to the early thirteenth century, Jayavarman VII, one of the most powerful rulers of the Angkorian Empire, extensively reorganised the royal capital and constructed major temples across his territories. Among these was the Buddhist temple of Banteay Chhmar, located approximately 110 km northwest of Angkor. Its extensive gallery walls were decorated with narrative bas-reliefs depicting military campaigns, religious subjects, courtly activities, and scenes of everyday life. These reliefs constitute important historical evidence for understanding the political circumstances surrounding Jayavarman VII’s rise to power and the representation of royal authority. Following centuries of deterioration and neglect, much of the stone-built temple collapsed, including extensive sections of its sculpted galleries. Recent reconstruction work has made additional relief surfaces accessible for direct observation, providing new opportunities to reconsider previous historical interpretations.

    Since 2024, a comprehensive digital documentation project has been conducted at the temple, covering the central sanctuary and surrounding archaeological remains. Terrestrial laser scanning, mobile LiDAR, and photogrammetry have been employed according to the characteristics of the structures being recorded. For the approximately 700 m of gallery reliefs, reaching around 3 m in height, high-resolution orthophotographs were generated through close-range photogrammetry to support detailed recording and analysis. Using these images, a web-based viewing system incorporating multiresolution image tiling was developed to facilitate seamless examination from the overall gallery composition to individual sculptural details. The system integrates orthophotographs, interpretive line drawings, and annotations concerning individual figures, narrative scenes, and historical sources, allowing users to explore the reliefs freely at different scales. The line drawings were produced with AI assistance to distinguish, to a certain extent, sculpted contours, joints between stone blocks, deterioration, and physical damage.

    This paper presents a new interpretation of the continuous narrative reliefs in the recently reconstructed western wing of the northern gallery, developed through analysis using this digital viewing system. Inscription K.227 from Banteay Chhmar records episodes of military conflict with the neighbouring kingdom of Champa before Jayavarman VII’s accession. Its latter section describes a campaign into Champa led by Prince Śrīndrakumāra, including the capture of a mountain fortress, an ambush during the return journey, the encirclement of the prince by Cham forces, and the sacrifice of two military commanders who fought to protect him. The reliefs in the western wing of the northern gallery are interpreted as presenting a sequence of scenes progressing from west to east that corresponds to these events. Particularly noteworthy are two elephant-mounted figures moving in the opposite direction to retreating infantry, as well as high-ranking figures apparently under attack or wounded. These representations provide compelling visual parallels with the events described in the inscription.

    The newly documented reliefs in reconstructed sections of the galleries offer further opportunities to reassess earlier interpretations and develop new historical understandings. This study demonstrates that integrated digital documentation and multiresolution visualisation facilitate continuous examination of fragmented narratives, interdisciplinary and systematic analysis in conjunction with epigraphic research, and the presentation and sharing of renewed interpretations, thereby contributing to advances in historical understanding.

  28. A Study on the Potential for 3D Documentation, Description, and Analysis of Temporal Changes of Culture Heritage

    Xuezhen Li, Tokai University, Japan · Paper ID 188

    Abstract

    This study proposes a 3D reconstruction method focused on the historical changes in cultural heritage and historic buildings. The value of cultural heritage is reflected not only in the architectural states that have been preserved to the present day, but also in changes, such as damage, alterations, additions, and repairs, that the structures have undergone during their long history of use. These traces left over time constitute part of the building’s history. Therefore, this study does not merely focus on the 3D preservation of a building’s current state, but rather treats changes in the building itself as the subject of documentation and presentation. To address the limited number of historical images and incomplete camera angles, this study does not attempt to reconstruct complete buildings from various historical periods. Instead, it records the current building based on the camera locations and perspectives of existing historical materials, identifies the parts that have changed by comparing images from different periods, and uses AI technology for single-image 3D reconstruction to assist in reconstructing partial past states based on a small amount of image data. Finally, spatial correspondences are established between the partial reconstruction results from different periods and different parts of the building and the complete 3D model of the current building, thereby presenting the partial changes in cultural heritage and historic buildings from the past to the present based on limited historical data.

  29. Utilising Gaussian Splatting as a new framework for digital preservation of stained glass

    Adrianna Polcyn, Visual and Technical CGI Artist, Scotland · Paper ID 110

    Abstract

    3D scanning has served as a vital tool for documenting, communicating, and preserving cultural heritage in the era of computer graphics technologies. However, traditional 3D scanning and capture techniques, such as photogrammetry, consistently fail when applied to complex optical features. When working with materials such as glass and lead, photogrammetry can often produce uneven, missing, or severely distorted geometry, compromising accuracy and undermining authentic interpretation. This ongoing research explores the integration of radiance field technology - specifically 3D Gaussian Splatting - as a transformative approach to digital heritage documentation. Unlike traditional mesh- and texture-based methods, Gaussian Splatting allows for accurately reconstructing thin, translucent, and transparent surfaces. Its advantage is also the ability to capture view-dependent properties, such as light reflections and refractions. This capability is instrumental for preserving the physical and artistic features inherent to stained glass designs. Furthermore, the resulting high-fidelity models can be leveraged for precise spatial measurements and non-invasive condition monitoring to track structural damage, material build-up, and wear over time. This study presents recent innovations within the 3D scanning field, addresses the technical bottlenecks of previously established workflows, and proposes a treamlined implementation pipeline optimised for conservation practitioners and further research into this field.

  30. 12:00–13:00

    Lunch Break

  31. 13:00–14:30

    Session 4: The Intangible Heritage in the digital Era

  32. Ruralities in the Archive: Community Participation and Digital Mediation of Rural Heritage

    Diana Mendes, Universidade do Minho - Lab2PT, Portugal · Paper ID 59

    Abstract

    This paper presents Ruralidades, a digital archive dedicated to the preservation and dissemination of the rural heritage of the municipality of Vila Verde, Braga, Portugal, across the twentieth century. Developed within a doctoral research project in History at the University of Minho, funded by the Portuguese Fundation for Science and Tecnology, and carried out in partnership with the House of Knowledge in Vila Verde, the project rests on a participatory, action-research methodology with the local community and municipal authorities. The Ruralidades archive gathers oral testimonies, photographs, and video around the concept of a "Living Library of the Rural", with guided visits, portraits of rural life, creative video, and a photographic collection, complemented by an interactive cultural map open to community-suggested contributions and a dedicated games section for heritage dissemination. By crossing oral memory with documentary sources, the project recovers domestic and agricultural roles, and the gender relations that structured the rural daily life of the community, which formal archives rarely register because they privilege written, official, and public record.

    Beyond preservation, the paper argues that public accessibility returns to the community a fuller reading of its own history. The paper discusses the methodological and technical choices underlying the creation of a digital platform on behalf of heritage and reflects on the urgency of collecting living memory before its loss.

  33. LITERATE: An AI-Enabled Participatory Framework for Safeguarding and Promoting Intangible Cultural Heritage

    Athina Grammatikopoulou, University of the Aegean, Greece · Paper ID 145

    Abstract

    The concept of Intangible Cultural Heritage (ICH) is defined in the UNESCO 2003 Convention as encompassing living expressions, knowledge, skills and practices through which communities maintain, transmit and continuously reshape their cultural identities. Over two decades, ICH has gained recognition for its contribution to social cohesion and sustainable development, particularly in tourism and education. Despite growing recognition, ICH is poorly documented digitally: searchculture.gr, the Greek cultural aggregator, returns only eight objects for the Mediterranean diet and mastic cultivation, two early UNESCO-inscribed elements. In addition to the limited findability and accessibility of ICH in national digital heritage infrastructures, communities of practice and bearers of ICH often have limited participation in digital documentation processes and are rarely engaged through systematic and reusable participatory research methodologies.

    Recent advances in Artificial Intelligence (AI), particularly Large Language Models (LLMs), Large Multimodal Models (LMMs) and Retrieval-Augmented Generation (RAG), offer new opportunities for organising, retrieving and accessing heterogeneous cultural resources. Building on these developments, the LITERATE project (IntelLigent InTangiblE CultuRAl HeriTage Experiences), co-funded by the European Union and the Greek State under the RESEARCH–INNOVATE programme (project code: ΕΚΠΑΡ02-0074153), proposes an integrated framework combining participatory research, multimedia knowledge management and AI-enabled applications to support the documentation and promotion of Greek ICH.

    Central to this approach is the development of a participatory methodology that actively involves local communities, cultural stakeholders and end users in documenting and enriching cultural heritage content, as well as in the participatory design and evaluation of digital experiences. Community-generated knowledge will be integrated with cultural collections through a dedicated crowdsourcing platform and a multimedia ICH repository supporting text, images, audio, video and associated metadata. The proposed infrastructure will incorporate mechanisms for data integration, harmonisation and quality control, alongside vector representations enabling RAG-based information retrieval.

    Building on this infrastructure, LITERATE aims to develop a bilingual (Greek/English) multimodal conversational assistant supporting voice interaction, multimedia queries and information retrieval from the ICH repository and external online sources. Complementary digital applications, including interactive maps, personalised recommendations and educational games, will facilitate access to and engagement with cultural heritage among citizens, tourists and diverse user groups, including children, older adults and people with visual impairments.

    “The integrated system will be modular, scalable, and adaptable, offering standards-based reusable content that can be customised to different needs. It will be tested, optimised and evaluated through pilot activities involving the Thessaloniki Tourism Organisation, the Olympic Museum and Aptaliko musical community, alongside community members, tourists and other end users. The evaluation will examine the system's functionality, usability and potential social impact.

    This paper presents the objectives and proposed methodological and technological framework of LITERATE, alongside an examination of the current state of the art in participatory ICH documentation, multimedia knowledge management and AI-enabled cultural heritage applications. It further discusses the opportunities and challenges of integrating community participation with advanced AI technologies, including considerations of intellectual property rights, ethical data governance and community involvement in the management and reuse of cultural knowledge, to support more accessible, inclusive and sustainable engagement with intangible cultural heritage.

  34. Vocabulary and Context in LLM-Generated Knowledge Graphs of Intangible Cultural Heritage

    Hüseyin Özçınar, Pamukkale University, Turkey · Paper ID 222

    Abstract

    When a heritage narrative is converted into a knowledge graph, the names of practices and participants may survive while the relationships between them become harder to recover. Who teaches whom, the order of a performance, and the conditions under which a ritual takes place can matter as much as the entities themselves. Comparison with a curatorial vocabulary introduces another difficulty: some catalogue labels are not expressed in the narrative supplied to the model. We examine how these problems affect the evaluation of knowledge graphs extracted from UNESCO Intangible Cultural Heritage descriptions.

    The study uses 30 elements, equally divided between Euro-Mediterranean and other regions, with 361 curatorial concept links and 90 selected source propositions. Two API-based model configurations were evaluated under generic and clause-role prompts. An additional agent-based configuration provides an exploratory comparison; its different execution procedure prevents a direct ranking of model capabilities. The provisional, LLM-assisted reference distinguishes concepts explicitly stated in the descriptions, concepts supported through inference, and labels unsupported by the supplied text. Rule-based measures assess lexical cue retention, directed core relations, and the attachment of two contextual qualifiers per proposition.

    Under the provisional classification, 60 of 280 secondary concept links lack support in the public descriptions, compared with none of 81 primary links. For concepts classified as inferable, recall of canonical vocabulary labels remains below 5%, while source-cue scores range from 88% to 100%. These results illustrate a limitation of label-based evaluation: failure to generate a catalogue term need not mean that the model discarded the wording associated with it. Conversely, repeating that wording does not demonstrate correct interpretation or alignment with the catalogue. The unsupported-link counts therefore describe the current reference rather than independently verified properties of the corpus.

    Across configurations, 63–72% of the selected propositions receive rule-based credit for a directed core relation with correct polarity in a single triple. Requiring both contextual qualifiers in that triple lowers the score to 29–51%. For the API-based outputs, allowing qualifier matches across triples sharing endpoint cues raises scores to 63–71%. This change shows how strongly the result depends on the evaluation unit. It does not establish that dispersed qualifiers remain attached to the correct event: shared words may occur in separate relationships without preserving their original connection.

    The contribution is a framework for separating source support, vocabulary alignment, relation accuracy, and contextual attachment when evaluating heritage knowledge graphs. The findings remain preliminary: manual verification of the reference is incomplete, independent second-coder validation is unavailable, and lexical rules can accept semantically incorrect matches. Human assessment is needed to determine how often lower single-triple scores reflect omission, recoverable decomposition, or incorrect attachment. For heritage documentation, these distinctions matter because retaining recognisable terms alone does not establish that a graph preserves the relationships that give a practice its meaning.

  35. From Digital Documentation to a Memory Twin: An Integrated Methodology for the Monumental Cemetery of Perugia

    Eleonora Dottorini, University of Perugia, Italy · Paper ID 215

    Abstract

    Historic cemeteries are complex cultural landscapes where tangible cultural heritage – architecture, sculpture and works of art – and vegetation intertwine with the anthropological and cultural significance of commemorative practices, individual biographies and collective memory. Digital surveying technologies now enable increasingly detailed documentation of the physical characteristics of heritage assets. Yet, particularly when cultural heritage comprises multiple interconnected elements, geometric representation alone cannot convey the historical relationships and intangible meanings they embody. A central challenge is therefore to integrate spatial documentation with historical sources, records and narratives through processes of interpretation, making knowledge accessible to diverse audiences and relevant to heritage protection and care.

    This paper presents an integrated, replicable methodology developed through the case study of the Monumental Cemetery of Perugia, Italy. Grounded in the Memory Twin paradigm, the research moves beyond digital replication towards a knowledge system connecting monuments with documentary sources and the memories they preserve. The aim is to bring knowledge acquisition, data capture, documentation, interpretation and communication into a coherent workflow designed for progressive expansion through further research and acquisitions.

    The methodology operates at two complementary scales. At site level, mobile laser scanning and aerial photogrammetry produced an overall point cloud supporting the documentation and spatial analysis of the cemetery layout. At monument level, terrestrial photogrammetric surveys were undertaken on selected structures according to knowledge-based criteria. Historical, architectural and iconographic investigations, including the histories of the commissioning families, relate their material characteristics to bibliographic and archival evidence. An analysis of burial practices, architectural types and funerary symbolism also establishes descriptive categories through which to interpret the site's diversity.

    The resulting knowledge is organised around individual monuments by integrating three complementary information layers: data, comprising geometric representations, images and documentary resources; metadata, describing objects and resources, their contexts and provenance; and paradata, recording acquisition and processing procedures, methodological choices, interpretive processes and associated uncertainties. This approach preserves also the knowledge of how that information was produced and interpreted, establishing the basis for a documented, verifiable and progressively updatable system.

    A virtual tour was developed as the spatial interface to this knowledge system. An interactive site plan provides access to 360-degree panoramas, from which users can consult monument records containing historical texts, photographs, audio content and detailed 3D models. These records are also arranged into thematic itineraries offering different perspectives on the heritage for visitors, local communities, researchers, technical professionals and public administrators. Among them, Atlante della Memoria introduces a participatory dimension, inviting citizens to share stories, photographs and other materials concerning people buried in the cemetery, thereby complementing historical and artistic documentation with individual and collective memories.

    The result is a replicable workflow linking site-wide documentation and monument-level investigation to the organisation and communication of knowledge within an accessible, extensible digital environment. The virtual tour thus serves as the interface to a Memory Twin in which data, documentary sources and memories can be continuously integrated. This approach offers a methodology adaptable to other funerary heritage contexts, connecting research, participation and the stewardship of places of memory.

  36. Bridging memory and technology: participatory digital models for heritage recovery and resilience

    Rafael Ramirez Eudave, Faculty of Civil Engineering and Geosciences, Delft University of Technology, Netherlands · Paper ID 139

    Abstract

    Loss of cultural heritage due to unforeseen or unavoidable circumstances has been discussed and researched in recent decades. While most efforts have focused on the material preservation of cultural assets, the intangible values that link them to their holding communities have only been considered an integral part of heritage preservation very recently. Nowadays, digital technologies have proven critical for the preservation and restoration of cultural assets. This research explores the capabilities of digital technologies in aiding the safeguarding and restoration of aspects such as the sense of place and belonging tied to cultural objects and events. Specifically, this paper presents some theoretical foundations for a strategic plan to enrich existing techniques for documenting cultural assets with information that helps contextualise objects as elements of a place and society. To achieve this, concepts such as genius loci (intangible values of the place), cultural heritage at risk, and participatory Geographic Information Systems (GIS) are explored to investigate how digital environments may reunite what has been dispersed or lost due to unforeseen or undesired actions, such as tourism, war, or natural hazards. Here, we propose leveraging digital approaches to construct a digital twin of a local genius loci through participatory processes (bridging citizens and heritage) alongside institutions such as museums and universities. We hypothesise that more holistic and comprehensive digital models of cultural heritage, designed to represent the sense of place as perceived by its communities, may be crucial for post-disaster recovery and improving resilience in risk scenarios. Moreover, such an approach could support the interpretation of complex cultural landscapes that exhibit multiple palimpsests across temporal and spatial scales. The paper concludes by outlining a roadmap to guide a mixed-method approach for documenting both the tangible and intangible values of cultural heritage, integrating expert knowledge, state-of-the-art surveying techniques, and citizen participation.

  37. From Digital Twin to Memory Twin: Integrating Tangible and Intangible Heritage in the Digital Reconstruction of a Sacred Cultural Landscape

    Lana Skender, Academy of Arts and Culture in Osijek, Croatia · Paper ID 130

    Abstract

    Digital heritage documentation has traditionally focused on recording, visualising, and spatially representing physical objects, often leaving their intangible meanings and relationships with living communities insufficiently explored. The emerging concept of the Memory Twin offers an opportunity to reconsider digital heritage representation by integrating material evidence with cultural memory, oral histories, social practices, and historically situated experiences. This paper explores the transition from object-centred digital documentation towards memory-oriented interpretation through the digital reconstruction of a rural sacred cultural landscape in eastern Croatia.

    The research builds on the project From Votive Offering to Heritage: The Saints’ Guard Between Past and Present, developed and awarded within the European Heritage Days Stories 2026 programme. The project documents a dispersed network of roadside shrines, crosses, chapels, and devotional sculptures in the parishes of Bebrina, Dubočac, and Slavonski Kobaš, originating from the early 18th century onwards. Historically, these monuments marked village boundaries, roads, and processional routes, forming a symbolic protective network known as the Saints’ Guard. Beyond their physical presence, they embody votive intentions, family histories, religious traditions, political ideologies, and communal practices that have shaped the region's cultural identity.

    The theoretical framework combines cultural landscape studies with theories of cultural memory developed by Jan and Aleida Assmann, Laurajane Smith’s understanding of heritage as a social process, and the UNESCO Convention for the Safeguarding of the Intangible Cultural Heritage (2003). These perspectives provide a foundation for interpreting sacred landscapes as dynamic relationships between material places, remembered experiences, and living traditions rather than static collections of monuments.

    The study adopts a qualitative case-study methodology integrating historical and art-historical research, spatial analysis, oral history, ethnographic interpretation, and reflexive thematic analysis. Historical maps, archival documents, field photographs, and geolocated monument records form the basis for reconstructing the sacred landscape's material and spatial organisation. Recorded interviews with local residents, particularly elderly women preserving memories of the monuments’ construction, votive origins, and religious processions, reveal additional layers of cultural meaning. Testimonies from younger generations concerning their participation in rituals offer opportunities to examine intergenerational transmission and transformations of cultural memory. The study envisages expert consultation with an ethnologist to contextualise these practices.

    Following Braun and Clarke’s approach, the interviews will be analysed thematically, focusing on votive meanings, family memory, ritual continuity, intergenerational transmission, and spatial relationships. Oral accounts will be compared with historical documentation, distinguishing verified evidence from inherited narratives and personal recollections while recognising their different interpretative values.

    The research addresses three questions: How can historical and spatial documentation contribute to reconstructing a sacred cultural landscape? How do oral histories and living rituals reveal intangible dimensions of its material heritage? How can integrating these sources support a transition towards memory-oriented digital heritage representation?

    Using StoryMapJS, the study explores how geolocated monuments can function as interpretative nodes connecting historical evidence, personal testimonies, and living traditions. Rather than claiming to construct a fully operational Digital Twin or Memory Twin, the research investigates a conceptual and methodological transition towards contextualised, memory-oriented digital representations.

    The anticipated contribution is an integrated approach to documenting sacred cultural landscapes that preserves relationships between physical monuments, historical transformations, and community memory, thereby expanding digital heritage interpretation beyond material reconstruction.

  38. From Digital Twin to MemoryTwin: Baghdad Cinematic Cut as a Method for Documenting Intangible Urban Memory in baghdad downtown

    Muna Baldawi, Al-Nahrain University, Iraq · Paper ID 95

    Abstract

    This research establishes a profound connection between the historical experiences of old Baghdad and an imagined vision — reminiscent of a scene from a film or a novel — of the city as it was first conceived: a "Dream Baghdad." It bridges the gap between an analytical study of Baghdad's urban layout and the lived reality of its social memory, explored through a spatial examination of its alleys, streets, and landmarks.

    While EuroMed's 20th anniversary proposes the shift from Digital Twin to MemoryTwin — from representing what heritage is to preserving what it has experienced and what it means — this paper proposes a third stage: Dream Baghdad, the city as envisioned by its admirers, including its architects, crafters, and citizens.

    The methodology is anchored in the notion of "Urban Cinematography," which frames the city as a cinematic scene. The research identifies the most crowded and frequently used urban locations as cinematic hotspots — these are the very areas that need intervention to represent what we call "Spatial Cinematography." Al-Mutanabbi Street, destroyed in the 2007 bombing and revived , serves as the primary case study.

    Three interlinked models are developed:

    1. Physical Twin (Analytical Study): High-resolution photogrammetric survey of current Al-Mutanabbi Street and archival documentation of post-2007 destruction. This charts the development of old Baghdad through its history of conflicts, crises, and many rehabilitation initiatives — some well-planned, others less so.

    2. MemoryTwin (Lived Reality): This layer captures social memory through oral histories of booksellers, documentation of the intangible Friday book market, and cinematic analysis of the street's transformation. The diptych visualization (2007 desaturated rubble with scattered books warm saturated recovery) demonstrates the shift from erasure to resilience. Each book stall is linked via a Knowledge Graph to stories, Iraqi films shot in the alleys, and literary references.

    3. Dream Baghdad (Future Vision): The final layer uses cutting-edge digital tools — XR, persistent virtual worlds, and AI-assisted generative design — to realize the city as imagined by its citizens. It is not a nostalgic reconstruction, but a near-future vision where the crowded cinematic locations identified earlier are enhanced as spaces of cultural encounter, where the dream of Baghdad's architects and craftsmen becomes navigable.

    Results show that urban recovery in old Baghdad cannot be achieved through physical restoration alone. The transition from MemoryTwin to Dream Baghdad requires re-activating spatial cinematography — the way people see, film, and narrate the alleys. The integrated model provides a circular economy of knowledge: one dataset serves structural analysis, memory preservation, and speculative design for sustainable tourism and education.

  39. Translating Craft Knowledge for Generative AI in Intercultural Fashion Design

    Yan Mao, Wuhan College of Communication, China · Paper ID 64

    Abstract

    Generative AI can expand the creative use of digital fashion heritage, but culturally informed design requires more than visual resemblance. A woven appearance may result from printing, textile construction or interlaced volume, each implying a different relationship to craft. This paper examines how comparative craft knowledge can be translated into design requirements that guide AI-assisted generation and explain the cultural reasoning behind a new garment. The study develops a conceptual framework through comparative artefact analysis. It takes as its starting point the November 2025 China–Cyprus fashion exchange in Nicosia, where Chinese and Cypriot designs were presented and a cooperation memorandum was signed. Event photographs, cooperation documents and visiting-scholar records are examined alongside designer accounts, museum records and official heritage inventories. The comparison focuses on Shandong Lujin handwoven textiles used in mamianqun, a form of pleated wrap skirt, and on basketry-inspired fashion by Pantelis Panteli and Livadia basket weaver Pavlina Loizou. These examples suggest two complementary approaches to heritage translation: reinterpreting an inherited garment form for a new cultural encounter, and adapting craft construction to wearable form. The framework connects two layers. A reusable representation links materials, operations, forms, meanings, contexts and contributors, retaining source identifiers and distinguishing documented accounts, visual observations and design interpretations. A project-specific transformation record specifies the source relation, the relation to retain, the proposed transformation, the scope for variation, a review criterion and an interpretive explanation. Four operations connect these layers: retaining, relocating, reinterpreting and juxtaposing. In the proposed workflow, AI retrieves craft relations, develops briefs, generates visual proposals, compares them with explicit criteria and records maker-approved revisions. The record also supports multilingual exhibition narratives that explain the design decisions. A worked scenario specifies three requirements for a mamianqun ensemble: an unobstructed front panel, a three-dimensional basketry-inspired component and a removable addition. Three hypothetical alternatives illustrate the resulting design reasoning. A printed shoulder piece does not fulfil the construction requirement; an interlaced apron obscures the front panel; an interlaced shoulder component is selected for further material development. Additional scenarios address botanical composition and an interactive exhibition of design reasoning. The contribution is a structured connection between comparative craft interpretation and generative design. Building on knowledge-augmented design research, the framework explains how distinct traditions can contribute complementary constructive and symbolic roles within a shared brief. The analysis demonstrates the framework through explicit design decisions rather than an implemented generative system or a performance evaluation. Collaborative making and user studies are proposed to assess its practical value for creative, curatorial and educational work.

  40. Preserving Creative Causation in Born-Digital Human–AI Literature: A Provenance Framework Beyond Prompt Archives

    Navi Musaget, FFRI-NM — Foundation for Relational Inquiry, USA · Paper ID 94

    Abstract

    Born-digital literature created through human–AI collaboration challenges cultural-heritage preservation. A finished manuscript shows what survived; prompt archives may document exchanges without explaining why a narrative changed, which alternatives were rejected, or how decisions carried across model changes. Machine-generated word counts measure textual quantity, not creative causation. How can archives preserve intelligible evidence of creative development without exhaustive capture of private interactions?

    This paper presents a conceptual design analysis of two related documentary sources: IIPSEON, a research line with documented protocols for causal and interpretive continuity across changing AI models, and the pre-launch MIELOGOS Literary Prize, an institutional framework separating creative-process eligibility from literary evaluation. Documents are selected for explicit treatment of causal inheritance, decision authority, eligibility evidence, or confidentiality. They are related design sources, not a representative sample of heritage institutions. Their provisions are compared through five questions: what changed; which interactions preceded it; which decisions mattered; what supports an inference of influence; and who may inspect or assess the record. A fictional multi-model revision scenario illustrates the design, not its operational effectiveness.

    The resulting selective provenance framework has five layers. The work layer records artifacts, versions, and transformations. The interaction layer identifies selected human–AI exchanges and computational-model changes. The decision layer records consequential choices: acceptance, rejection, revision, or refusal that materially redirects text, structure, interpretation, or later creative decisions. The causal-link layer separates verifiable records of events from the more contestable inference that one event influenced another. A dated proposal and revision can corroborate a sequence; a recorded authorial rationale may support, but cannot conclusively prove, causation. The access-and-authority layer identifies sources, permissions, and the human author or curator responsible for assigning an evidence grade, with the justification and disagreements retained. Significance and retention judgments remain contextual and are documented through selection notes, including known omissions.

    The proposal complements existing provenance approaches rather than claiming implemented compatibility: W3C PROV models entities, activities, and agents; CIDOC CRM supports event-centred cultural-heritage representation. The proposed contribution concerns graded causal claims, meaningful decisions, selective retention, and access constraints. To support interpretation across changing communities, records should preserve contemporaneous rationales, contextual vocabularies, and the reasons for selection, not only graph links.

    In the illustrative case, a short model criticism prompts a substantial narrative revision while longer machine-generated passages are rejected. The record can represent this asymmetry without invented contribution percentages. Recording computational participation does not settle AI agency, subjectivity, or authorship.

    The contribution is an archival design proposal, not a tested system. Interoperability, consistency of evidence grades across curators, selective-record bias, privacy, creator burden, and long-term usability require practical evaluation.

  41. 14:30–14:50

    Coffee Break

  42. 14:50–17:00

    Session 5: Digital Documentation and Knowledge extraction

  43. A Blueprint for the Digital Representation of Historical Books and Manuscripts: Integrating Semiotic and Material Evidence

    Xenophon Zabulis, FORTH, Greece · Paper ID 124

    Abstract

    Historical books and manuscripts are simultaneously carriers of textual and graphic information and physical artefacts whose materials, construction, alterations and traces of use provide evidence of their production and history. Digital documentation has traditionally privileged their informational content, transforming them into collections of page images, transcriptions and searchable records. However, their material properties contain complementary historical evidence that cannot be recovered from textual content alone.

    Representing historical books and manuscripts as cultural heritage entities within digital knowledge infrastructures therefore requires methods that account for both their tangible and intangible dimensions. The challenge extends beyond digitisation to the acquisition, interpretation, integration and preservation of heterogeneous evidence concerning the same historical object, while retaining the distinctions between different forms of observation, computation and scholarly interpretation.

    This paper presents the methodological and technological blueprint developed for the Horizon Europe ARXIVE project. It establishes two complementary analytical paths. The semiotic path addresses textual and graphic content through document layout analysis, text recognition, structured encoding, semantic annotation and knowledge extraction. The material path addresses the physical artefact through imaging, geometric and structural documentation, material characterisation, and the study of bindings, construction, condition and traces of intervention. Each path generates distinct observations, representations and scholarly interpretations. Their integration requires shared referents and explicit relationships without reducing the two forms of evidence to a single descriptive model.

    Metadata and paradata provide essential mechanisms for this integration. Descriptive, structural, technical and semantic metadata identify heritage entities, organise their components and representations, and support discovery and interoperability. Paradata records the processes through which digital representations and analytical results are produced, including acquisition conditions, processing methods, computational parameters, transformations, uncertainties and validation decisions. Together with persistent identifiers and provenance records, these mechanisms establish traceable relationships between physical objects, digital resources, analytical results and scholarly claims. They also support the reassessment of interpretations as new evidence becomes available.

    The proposed blueprint organises these capabilities into an evidence-preserving workflow spanning acquisition and processing, semiotic and material analysis, semantic integration, knowledge management and presentation. It draws upon a review of relevant technologies, methods and standards, while incorporating professional workflows, human validation, institutional requirements, long-term preservation, and applicable legal and policy frameworks. The resulting synthesis identifies technological capabilities, limitations and dependencies, and translates them into architectural principles and recommendations for subsequent ARXIVE development.

    The proposed architecture does not seek to produce a definitive digital replica of a historical book. Instead, it establishes the methodological basis for interoperable, evolving representations in which textual and material evidence remain distinguishable, mutually accessible and traceable to their origins. Its objective is to support the integration of historical books and manuscripts into digital heritage infrastructures without losing the physical, intellectual and historical evidence through which they are understood.

  44. Traditional Papermaking Techniques and Their Relevance to the Preservation and Conservation of Buddhist Manuscripts

    Tanya Bodh, Panjab University, India · Paper ID 190

    Abstract

    Buddhist manuscripts are important parts of cultural heritage because they retain religious teachings, philosophical ideas, literature and historical knowledge. Their preservation depends not only on proper storage and handling but also on understanding the materials used to produce them. Traditional papermaking in Tibet and other Himalayan regions has relied on locally available plant fibres as well as traditional paper preparation and treatment procedures. Studying such materials and techniques can help improve our understanding of historical manuscripts and their preservation and conservation.

    This paper will examine traditional papermaking materials and techniques and explore their relevance to the preservation and conservation of Buddhist manuscripts. It adopts a qualitative, literature-based approach using published research on traditional papermaking, historical Tibetan paper and manuscript conservation. The literature is reviewed thematically, focusing on plant materials, papermaking processes, paper properties and their possible applications in conservation. The study is based on previously published research and does not involve new laboratory experiments or primary data collection.

    The reviewed literature highlights the use of plant fibres from the Thymelaeaceae familyincluding Stellerachamaejasme and species of Daphne and Edgeworthia in discussions of traditional Tibetan and Himalayan papermaking. The preparation of these plants involves processes such as fibre extraction, boiling, beating and sheet formation. These stages can influence the physical characteristics of the finished paper.Scientific analysis of paper fragments from Samye Monastery, dated to the twelfth and thirteenth centuries revealed plant fibres, wheat starch, milk-derived proteins and calcium carbonate providing insights into the materials and treatments used in historical Tibetan paper production. Published conservation research also describes the use of paper prepared according to Himalayan papermaking traditions in the repair of fragmented Buddhist manuscripts.

    The review suggests that the selection and preparation of plant fibres, along with paper treatments, are important for understanding the physical properties of historical paper and selecting suitable conservation materials. However, the preservation of manuscripts also depends on storage conditions, handling, environmental exposure and previous repairs. The study further highlights the value of documenting traditional papermaking knowledge, including the materials, techniques and skills maintained by regional communities. Such documentation can contribute to safeguarding intangible cultural heritage while supporting the preservation of tangible heritage, including Buddhist manuscripts. Further comparative research into historical paper composition, material properties and traditional production methods is needed to strengthen the evidence available for conservation decisions.

  45. Knowledge Fields: Reading Paradata and Object Biography on the Surface of a Memory Twin

    Evgeny Kalachikhin, Film University Babelsberg, Germany · Paper ID 213

    Abstract

    Photogrammetric models present every part of their surface with the same visual authority, whether it was captured, repaired, inferred or synthesised. Paradata, as established by the London Charter and the Seville Principles and refined in information studies, document the processes and decisions behind such data. Knowledge about the object itself, its biography, condition and meaning, lies outside this definition; the Memory Twin brings both together by integrating data, metadata, paradata and intangible values. In practice these layers remain in separate records or reach the model as a single uncertainty scale.

    We present Knowledge Field, a method that anchors such a layered knowledge record to every vertex of a 3D model. It combines four families, acquisition metadata, process paradata, interpretive paradata such as confidence and review, and object knowledge such as biography and condition, and keeps the term paradata for the process layers only. The record is stored as glTF vertex attributes linked to CIDOC CRM-oriented event records, separate from any visual encoding, so that it travels between tools: the prototype runs in a web browser, the same attributes already import into Blender, and implementation in 3D packages and real-time engines such as Unreal Engine is the next step. Three dimensions are projected into the perceptual OKLCH colour space, and switching projections tours the data space. Time is encoded geometrically: each documented event emits a shell of a glowing field from the surface points it affected, at a distance that grows with its age, analogous to lookback time in astronomy. Dating uncertainty appears as blur, source strength as density, conflicting sources as split shells, and gaps in the record as empty space. A time cursor navigates the history of the object and of knowledge about it, and a continuous morph moves the surface between its physical shape and its position in data space. Estimated, inferred and synthetic values carry a constant visual mark regardless of the projection, and an optical channel makes the record decodable from screen frames.

    The method is demonstrated on a smartphone scan of a bronze fountain figure in Munich, cast in 1963 after Andrea del Verrocchio's Putto with Dolphin (c. 1470), whose Florentine original was replaced by a copy in 1959. As a copy of a copy whose model is undocumented, the figure makes the trajectory of a reproduction legible in the sense of Latour and Lowe. Every surface point carries four times: the form of the Renaissance design, the matter of the 1963 cast, the surface accumulated since, and the data produced since 2023. Process paradata are measured from the scan files, object knowledge comes from documented sources, and archival condition records are synthesised and labelled as such. A first prototype recovered all values from its rendered frames with a maximum error of 0.5/255.

    We argue that spatially anchored knowledge records offer a practical route from the Memory Twin as a concept to an interface in which the epistemic status, history and lineage of a digital heritage object become readable on its surface.

  46. Create Advanced Digital Cultural Heritage Twins Using High-Level Knowledge Representation Techniques

    Gian Piero Zarri, STIH Laboratory, Sorbonne University, Paris, France · Paper ID 62

    Abstract

    This paper presents a synthesis of some recent experiments related with the implementation of innovative proposals to generate the "digitalized images" of any possible Cultural Heritage (CH) entites. We suggest, then, to represent under "Digital Cultural Heritage Form" the set of those intangible features corresponding to the symbolic/semantic/conceptual properties of the original CH entities - often neglected in the digitization processes. This approach will imply the use of Knowledge Representation techniques more advanced with respect to those used to encode the physical counterparts of our Twins, given the need to represent in digital format immaterial notions like feeling, sentiments, emotions, mutual relationships, etc. For the experiments mentioned above, we have utilized a high-level, high-expressiveness tool as NKRL, the Narrative Knowledge Representation Language, which corresponds also to a fully implemented Computer Sceince environment. Examples of Digital CH Twins of different complexity corresponding to concrete CH situations are presented in the paper.

  47. Beyond Plausibility: A Provenance-Aware Knowledge Graph Framework for Evidence-Grounded AI Reconstruction of Underwater Cultural Heritage

    Menna Salah Afiffi, Ain Shams University, Egypt · Paper ID 73

    Abstract

    Artificial intelligence (AI) is expanding the possibilities of three-dimensional (3D) reconstruction and interpretation of underwater cultural heritage. Yet visually plausible reconstructions may obscure the distinction between documented archaeological evidence, scholarly interpretation, computational inference, and hypothetical completion. This raises a central question: how can the evidential basis of AI-assisted reconstructions be made transparent, traceable, and open to scholarly scrutiny? This paper proposes a provenance-aware knowledge graph framework for evidence-grounded AI-assisted reconstruction, using selected archaeological elements from Thonis-Heracleion, the submerged ancient Egyptian port city in the western Nile Delta, as a case study. Rather than reconstructing the entire site, the proposed study focuses on three to five elements for which relevant published documentation can be identified. It combines an epistemic classification model informed by the Extended Matrix approach with CIDOC CRM-based semantic representation to connect archaeological objects, documentary sources, scholarly claims, interpretative decisions, computational operations, and reconstructed 3D components. The framework distinguishes four epistemic categories: E1, documented archaeological evidence; E2, scholarly interpretation; E3, computational inference; and E4, hypothetical reconstruction. These categories describe the epistemic status of individual claims and model components, not confidence levels. Uncertainty is recorded separately through source references, explicit annotations, and, where appropriate, confidence assessments. The resulting knowledge graph is intended to preserve provenance across the reconstruction workflow, allowing users to trace selected digital components to their supporting evidence and identify where interpretative or computational assumptions enter the process. The proposed methodology comprises evidence inventory and quality assessment, semantic knowledge modelling, provenance tracking, and evidence-aware 3D representation. AI-assisted interpretation and reconstruction are treated as bounded, reviewable operations rather than autonomous sources of archaeological truth. A small-scale prototype will be evaluated through traceability tests, identification of unsupported claims, and expert review of the consistency between selected reconstructed components and their documented evidential basis. Particular attention is given to epistemic drift: the process through which interpretative or AI-generated hypotheses may become detached from their original evidential status and be perceived as established archaeological facts. By operationalizing reflexive authenticity through a traceable knowledge infrastructure, this research aims to move beyond visual plausibility towards accountable digital reconstruction. Its anticipated contribution is a reusable methodological framework for integrating provenance, uncertainty, semantic interoperability, and AI-assisted interpretation into digital cultural heritage workflows, with potential applications beyond underwater archaeology.

  48. What a Knowledge Graph Cannot Say It Has Not Looked For: Declared Absence as a Typed Entry in Cultural Heritage Provenance

    Christopher Banks, Objects of Affection Collection LLC, USA · Paper ID 79

    Abstract

    Cultural heritage knowledge graphs inherit the open world assumption, under which an absent statement is not a negative statement. A provenance graph therefore cannot distinguish three conditions that are different facts about the world: a search was made and returned nothing; the document was never created or never kept; and nobody has yet looked. All three serialize as the same missing property, and they reach the reader as the same silence.

    The consequence is practical rather than philosophical. A research act that returned nothing is itself knowledge, and it is the one result the record has no way to carry, so it is repeated by the next researcher. A record examined and found clean is byte-identical to one never examined.

    This paper reports measurement across published provenance, then an instrument, then the instrument failing. Of ten major museums selected for open data availability, three publish provenance as machine readable data at all; the largest exposes it in neither its API nor its bulk dataset. Across the full corpus of one national gallery, n equals 81,543, approximately 0.18 per cent of records declare that a search was made and failed, while 2.75 per cent name an anonymized holder, making a withheld party roughly fifteen times commoner than a declared failed search. Across two Linked Art deployments, 120 objects, there are no typed custody events at all; in the reference implementation provenance is carried as a text object with a classification on it. Adopting a structured vocabulary made the prose addressable without making the provenance machine readable, which is a design position its architects state openly rather than a defect discovered here.

    Against that baseline we describe an openly licensed specification in which a gap is a typed entry rather than a blank, stating which kind of absence it is, why, who could still close it, and the date it was last checked. It was entered to the CIDOC CRM Special Interest Group as issue 723 and heard in October 2026; the group left it open and asked for further cases. Outside implementations number zero.

    The findings that matter most are against the instrument. Read strictly, its category for an absence that is recoverable but not yet obtained is very nearly unsatisfiable from record text, because records do not attach a documentary source to the gap they are silent about. A record built from a state land reform archive could declare no interval gap at all, because none of its events carries a date. And the authors produced a false declared absence in their own corpus three times, labeling an unperformed check as a performed one, which is precisely the confusion the specification exists to prevent.

    We offer a vocabulary, a measurement baseline, and the open problems.

  49. AESOP: Graph-Grounded Annotated Bibliographies from Reviewed CIDOC CRM Claims

    Lazaros Ioannidis, International Hellenic University, Greece · Paper ID 174

    Abstract

    Annotated bibliographies depend on statements that can be traced to the publications they summarise, attributed to their authors and compared across sources. These properties call for an explicit record of who asserted what, at the level of individual claims. We present AESOP, a system that grounds the generation of annotated bibliographies in a knowledge graph of reviewed scholarly claims about heritage entities, developed within the Horizon Europe project ARXIVE.

    In the sources pipeline, a language model extracts candidate claims from open-licence publications as instances of CIDOC CRM patterns, such as a production, a transfer of ownership or a carried text. Each claim carries exact quotations, a modality and an attribution to the author or to an authority the author reports. Deterministic code resolves the entities against a curated registry, validates every quotation and maps each claim to a proposition set of the Heritage Digital Twin Ontology (HDTO). A reviewer confirms the claim and includes it in the digital twin of the heritage entity concerned. Identifiers, RDF and SPARQL are generated exclusively by code.

    In the bibliography pipeline, a research topic is analysed into relations over the same claim types, with bound entities and open slots. Code compiles these relations into SPARQL queries over the reviewed claims, which return answers, rival claims from different publications, linked answers and event chains. The results steer a hybrid of lexical, vector and graph retrieval that selects the publications and passages for drafting. A language model drafts one annotation per publication, each representing its own source, and a verifier assesses every sentence against exact quotations from the cited passages.

    We describe the pipeline and its tunable parameters, including window size, schema strictness, claims per window, claim-type versions, retrieval limits and drafting checks. We report a preliminary benchmark of nine open-weight models and one reference model on a descriptive catalogue of late Byzantine Greek manuscripts at the Wellcome Collection, with each model tuned to its best configuration. On held-out catalogue entries, a 14-billion-parameter model on a 16 GB consumer GPU recovers 71% of the expected claims in mappable form, against 71-81% for the reference model, which in turn drafts annotations with fewer unsupported sentences. The expected claims await review by domain experts, and the figures are therefore preliminary. Future work comprises an expert-reviewed evaluation in the ARXIVE pilots, links to domain authorities, interval-based dates, SHACL shapes, and claim types for archaeological and stratigraphic documentation.

  50. Beyond Digital Representation: Knowledge Formalisation and Historical Interpretation in Architectural Heritage Research

    Luca Cei, UNIFE, Italy · Paper ID 195

    Abstract

    The extension of digital methods to architectural history has enlarged both the range of evidence available for investigation and the operations by which it is selected, compared and described. Three-dimensional acquisition, machine-learning procedures and information models, however, do not merely accelerate established research practices: the decisions through which material and documentary evidence is translated into computational objects also condition the questions that can be asked. The constructed character of data has long been discussed within digital humanities (Drucker 2011), as has the mediation of archaeological knowledge by digital infrastructures (Huggett 2015). In architectural conservation, related concerns arise from the need to maintain a critical relationship with historic fabric when its study increasingly relies on digital representations (Fiorani 2021).

    The paper draws on two doctoral investigations to consider different directions of this translation. The first examines automated segmentation and classification of three-dimensional acquisition of carved architectural elements. Although the models themselves already enabled more in-depth morphological comparisons within a still entirely manual, non-automated framework, the algorithms’ ability to recognise critically relevant distinctions remained constrained by the available datasets and the categories defined for training: the tests did not identify new groupings of established historical significance. The second investigation develops a protocol for organising historical-critical information in Heritage Building Information Models, starting from the case study of some Ferrara palaces, relating material evidence, documentary sources and hypotheses concerning architectural transformations through defined information units and controlled descriptors. It addresses the well known difficulty of accommodating historically heterogeneous and provisional knowledge within BIM structures (Della Torre and Pili 2020).

    Despite their different aims, both investigations translate complex evidence into discrete, operable categories, requiring decisions on comparable features, the identification and dating of architectural elements, and the treatment of conflicting or incomplete evidence. These choices precede computation but may be obscured by its results. A morphological class does not establish a historical relationship, nor does a model attribute explain the grounds for a dating or interpretation. Formalisation enables broader comparison and systematic enquiry, but may detach information from its context of production and present provisional conclusions as established facts.

    These issues also affect interoperability. Shared standards facilitate the exchange of three-dimensional resources, annotations, metadata and paradata, yet technical compatibility does not guarantee the preservation of historical meaning. Distinctions between observation, inference and interpretation must remain legible throughout data processing. Rather than relegating critical judgement to a final check, the paper considers its role from the outset in defining datasets, categories, relationships and validation criteria. Comparing the two investigations, it asks when digital formalisation can extend historical enquiry while keeping its interpretative premises open to scrutiny and revision.

    Short Bibliography

    Della Torre, Stefano; Pili, Alessandra (2020). Built Heritage Information Modelling/Management. Research Perspectives, in Digital Transformation of the Design, Construction and Management Processes of the Built Environment, Springer, Cham, pp. 231–241.

    Drucker, Johanna (2011). Humanities Approaches to Graphical Display, in “Digital Humanities Quarterly”, 5(1).

    Huggett, Jeremy (2015). A Manifesto for an Introspective Digital Archaeology, in “Open Archaeology”, 1(1), pp. 86–95.

    Fiorani, Donatella (2021). Conservazione vs Innovazione? Insegnare (ancora) restauro in architettura, in “ArcHistoR”, Extra 9, pp. 42–57.

  51. Unifying Tangible and Intangible Aspects of Natural Cultural Heritage

    Rafael Prezado, University of Évora, CIDEHUS, Portugal · Paper ID 50

    Abstract

    This paper presents an ontology model for natural cultural heritage sites. The model considers both tangible and intangible dimensions, connecting environment and geological characteristics with cultural significance, based on human stories and experiences. The ontology development is in pair with an eXtended Reality (XR) installation, and is based on Koli, a natural heritage site located in Finland.

  52. Beyond Archaeological Databases: Representing Chronological Evidence, Context, and Uncertainty in a Knowledge Graph for Northwest African Prehistory

    Aya Mohamed, Cairo University, Faculty of Archaeology, Egypt · Paper ID 175

    Abstract

    The growing availability of archaeological datasets offers new avenues for comparative research. However, assessing the support, context, or revision of archaeological claims remains challenging, especially in prehistoric archaeology. Here, site dates may not directly reflect human occupation, necessitating a combined evaluation of cultural attribution, stratigraphic association, and chronological reliability.

    This paper introduces a provenance-aware knowledge representation framework for integrating chronological and contextual evidence from Northwest Africa's Middle and Later Stone Age. It builds upon Boisard and Ben Arous's (2024) inventory of 993 sites and 1,152 dates across Morocco, Algeria, Tunisia, and Libya, which highlights both the potential of systematic data integration and the need to differentiate chronological evidence types.

    The framework moves beyond viewing sites and dates as isolated entries, modeling relationships between sites, contexts, dated materials, dating determinations, chronological intervals, cultural attributions, publications, and interpretative claims. Each claim is linked to its source, retaining information on dating methods, contextual associations, uncertainty, and evidential limitations from published records. This structure aims to preserve the distinction between observations and interpretations, rather than conflating them.

    Methodologically, the study outlines a workflow for mapping selected inventory records into a semantic model, guided by established cultural heritage knowledge representation principles. The model is evaluated using competency questions: Which dates directly link to human occupation evidence? What sources support specific chronological or cultural attributions? Where do alternative interpretations exist? Which records lack sufficient contextual information for reliable comparison? These questions determine if the model enhances transparent retrieval and cross-site analysis beyond conventional tabular representations.

    The goal is a domain-specific archaeological knowledge management approach that emphasizes provenance, contextual relationships, and chronometric data.

  53. Connecting Heritage Resilience Assessments with Supporting Evidence: A Conceptual Digital Decision-Support Design for Bagan, Myanmar

    Ei Thandar Kyaw, Kyushu University, Japan · Paper ID 219

    Abstract

    Heritage resilience assessments can help identify conservation needs, but their usefulness depends on how clearly users can understand the evidence and reasoning behind them. Overall scores may conceal specific weaknesses, while charts alone may provide little explanation of how assessments were reached. This paper proposes a conceptual digital decision-support design for Bagan, Myanmar, connecting resilience scores with supporting evidence and possible management responses. It is proposed as an initial component of the Southeast Asian Heritage Resilience Digital Observatory (SEAH-RDO).

    The design draws selectively on completed doctoral research comparing seven heritage cases in Southeast Asia. The underlying research combined field observations, questionnaire surveys, documentary review, and contextual stakeholder discussions. It developed a nine-dimension Heritage Resilience Matrix, a composite Heritage Resilience Index, and visual Resilience Profiles. The present paper focuses on Bagan to explore how these existing assessments could be made easier to interpret and trace within a digital interface.

    Bagan’s assessment illustrates the limitations of relying on an overall score. Its composite index indicates low-to-moderate resilience, while the dimension-level assessment identifies particularly weak tourism management and digital heritage management capacity alongside comparatively strong community participation. These findings describe the original study period and are treated as preliminary, interpretive assessments rather than current measurements or predictions.

    The proposed design comprises a site overview, a resilience dashboard, an evidence viewer, and a management-response view. All nine dimensions would remain visible to preserve the assessment’s broader context. Three dimensions-tourism management, digital heritage management, and community participation-would provide detailed examples of the connection between scores and evidence. Selecting a dimension would reveal its score, explanation, supporting sources, evidence dates, and limitations. The management-response view would present possible actions for consideration, clearly distinguished from documented findings.

    A structured data model would link site, assessment, evidence, source, and proposed-response records through unique identifiers. This would allow users to follow an assessment back to its supporting material and identify missing information. Public-facing records would use appropriately anonymised summaries and respect permissions for reuse. A proposed review process would allow heritage managers and community contributors to question interpretations and suggest updates.

    The contribution is a focused, research-informed design for making heritage resilience assessments more transparent and open to review. Future development would implement a small working prototype and examine record completeness, evidence traceability, and accurate retrieval of linked information. Subsequent stakeholder evaluation would assess interpretability and usefulness in discussing conservation needs. The paper does not claim a functioning platform or validated improvements in management outcomes. The Bagan illustration establishes a basis for testing the approach before considering its extension to other Southeast Asian heritage contexts.

  54. From Historical Records to Institutional Memory: A Knowledge Management Approach to University Library Heritage

    Somaye Sadat Akhshik, Kharazmi University, Iran · Paper ID 83

    Abstract

    University libraries preserve evidence of institutional development, yet their historical records are frequently dispersed across annual reports, correspondence, administrative regulations, photographs, and institutional publications. Such fragmentation complicates the reconstruction of institutional histories and limits the reuse of documentary heritage in research, education, and public engagement. This paper examines how documentary analysis and knowledge organization can provide a foundation for treating university library heritage as structured institutional memory.

    The study draws on a historical documentation project concerning Kharazmi University Library in Iran, tracing its development from its origins in 1918 through subsequent institutional transformations. The documentary corpus includes historical annual reports, administrative correspondence, library regulations, photographs, and records of organizational change. These sources have been examined to reconstruct significant events, identify institutional actors, and situate changes in library governance, collections, and services within their historical context.

    Building on this documentary work, the paper proposes a five-stage knowledge management framework: (1) source identification and authenticity assessment; (2) extraction of actors, events, decisions, and documentary relationships; (3) organization of contextual metadata and provenance; (4) chronological and thematic representation of institutional knowledge; and (5) dissemination through interpretive digital information services. The framework emphasizes traceability to primary sources, temporal context, and the distinction between documented facts and interpretive connections. It is a conceptual synthesis informed by a documented case, not an evaluation of a deployed knowledge graph or artificial intelligence system.

    The case illustrates the potential for historical library records to support contextualized digital exhibitions, navigable timelines, and future semantically enriched discovery services. By linking documentary evidence with knowledge management processes, the framework offers an approach to preserving not only archival objects but also the relationships and decisions that give them institutional meaning. Further implementation and user evaluation would be needed to establish technical feasibility, usability, and transferability across institutions.

    The contribution is a source-grounded, transferable conceptual model for academic libraries and university museums seeking to transform dispersed historical documentation into accessible, reusable institutional heritage knowledge.

  55. Digital Restitution: Digital Robe – A Case Study on Dispersed Textiles from Turfan and Dunhuang

    Christiane Isenberg, Hanseatic Blockchain Institute, Germany · Paper ID 143

    Abstract

    Textile heritage associated with Turfan and Dunhuang is dispersed across museum collections, complicating the reconstruction, documentation, and comparative study of historical garments. Fragmented object records and inconsistent cataloguing practices can obscure relationships between physical textile fragments, historical garments, and modern reconstruction decisions. This paper examines how semantic metadata, three-dimensional reconstruction, and blockchain-supported provenance can improve access to dispersed textile evidence without conflating digital representation with legal ownership or physical restitution.

    The Digital Robe project provides a case study of a digital garment reconstruction supported by a CIDOC Conceptual Reference Model (CIDOC CRM)-informed semantic metadata schema and a token-linked digital record. The research addresses three questions: how semantic metadata can represent relationships among dispersed textile fragments, historical garments, reconstruction activities, and custodial institutions; what evidentiary function blockchain-supported records provide and what their limitations are; and how museum professionals discuss the prototype's research potential and implications for digital restitution.

    The semantic architecture distinguishes physical textile fragments, documented or hypothesised historical garments, reconstruction activities, digital representations, attributed interpretations, and tokenised records. It draws on CIDOC CRM concepts to represent relationships while distinguishing recorded evidence from conjectural interpretations. In particular, the model provides a basis for documenting who attributes a relationship, which entities are involved, and what evidence or reasoning supports the attribution. This distinction is important because a visually coherent digital reconstruction can otherwise obscure uncertainty concerning the association of fragments with a proposed historical garment. The current prototype records uncertainty through qualified notes and documented modelling decisions rather than a formal uncertainty ontology.

    The implementation comprises one reconstructed robe, its associated semantic representation, and an ERC-721 record tested in connection with the Polygon Mumbai testnet and content-addressed digital assets. The reconstruction workflow combines digital acquisition, mesh preparation, material authoring, garment visualisation, and interactive presentation. The provenance mechanism associates the digital asset with metadata and a SHA-256 cryptographic digest, enabling comparison between a registered digital fingerprint and a subsequently retrieved file. The paper distinguishes these implemented prototype components from proposed developments, including broader cross-institutional integration and more formalised uncertainty representation.

    The prototype was presented during a professional workshop in Berlin attended by 32 museologists and archaeologists from Chinese national museums and heritage institutions. The session provided an exploratory setting for discussing digital reconstruction, semantic documentation, provenance, and the relationship between digital access and physical restitution. It was not a controlled experiment and does not establish statistically generalisable effects or causal improvements in cultural engagement.

    The principal contribution is a semantic framework for making reconstruction decisions, evidential relationships, and interpretative limitations more explicit and potentially queryable. A cryptographic hash can support digital-file integrity checking, but cannot independently establish historical accuracy, truthful provenance, consent, cultural authority, or ownership of a physical object. Digital restitution is therefore framed as discourse infrastructure that supports comparative research and informed discussion, rather than as a substitute for material return.

  56. The Israel Project: From Architectural Archives to a Repository of Possible Histories

    Michael Cidor, Bezalel Academy of Arts and Design, Israel · Paper ID 128

    Abstract

    Architectural archives are typically structured around the boundaries of collections, individual creators, and discrete documents. Although effective for preservation and retrieval, this structure often obscures the historical processes that connect buildings, people, institutions, places, and events across multiple archives. How might an architectural archive operate not only as a repository of evidence, but also as an evolving infrastructure for constructing and questioning historical narratives?

    This paper presents *The Israel Project: A Repository of Possible Histories*, an ongoing research and archival initiative at the Bezalel Academy of Arts and Design. Its heterogeneous corpus encompasses approximately 100,000 items from 27 metadata-rich architectural collections, some 60,000 additional materials organised under roughly 120 thematic “prisms” with uneven description, and approximately 10,000 student projects and associated research materials. The project seeks to connect these differently structured sources without erasing their provenance or forcing them into a uniform cataloguing hierarchy.

    We propose a deliberately lean, event-oriented knowledge framework that treats people, institutions, places, and time as common reference points, while describing historical activities through controlled, source-linked relational assertions. Rather than considering a building a single stable record, the model can represent its commissioning, financing, design alternatives, approval, construction, use, alteration, representation, and demolition as distinct, temporally situated actions. Assertions retain the documentary source, attributed roles, chronological precision, and uncertainty, making it possible to accommodate partial or contradictory evidence. Existing authority infrastructures and semantic standards, including national and international authority files and CIDOC CRM, provide points of alignment rather than a substitute for historical interpretation.

    The archive’s growth is driven not by a single collection-building programme, but by a cumulative body of diverse research and teaching activities concerned with architecture and planning in Israel. Each investigation encounters different sources, produces new descriptions and interpretations, and establishes connections that extend beyond the boundaries of individual collections or thematic categories. Rather than treating research as a secondary use of a pre-existing repository, the project considers it a primary mechanism of archival accumulation and knowledge production. This reciprocal relationship—in which research continuously expands and reorganises the archive, while the archive enables further research—also informs its information architecture. Thematic “prisms” operate as overlapping and revisable pathways through the corpus, allowing materials assembled in one research context to become discoverable and meaningful in many others.

    The paper discusses the transition from heterogeneous metadata and exploratory graph datasets toward an interoperable archival and research infrastructure, including a planned integration with the Alma digital repository and query-driven visual access. It argues that the value of a heritage knowledge graph lies not in producing one comprehensive historical account, but in making multiple evidence-based accounts inspectable, revisable, and open to further research and teaching. The Israel Project thus offers a methodological framework for treating architectural archives as dynamic repositories of possible histories.

Wednesday 4th November 2026

All times in CET · 100% online

08:00–09:30

Session 6: Beyond Digitisation: Unlocking the Power of Cultural Heritage - Use & Reuse - Part I

  1. R, Python, and the FOSS model for data accumulation, analytics, and distribution

    Taylor Rockhill, SOAS, University of London, United Kingdom · Paper ID 201

    Abstract

    As we continue to build databases to enhance research we need to be sure we're building these databases to be platform agnostic, and not beholden to private entities whose priorities may not have been disclosed, especially foreign private entities that may come under the influence of potentially belligerent foreign powers. One of the problems facing modern academia, especially outside the US, is the over-reliance on digital infrastructure controlled by US-based corporations subject to US laws and data norms. Recent events have made clear that reliance on US-controlled digital infrastructure represents a considerable security risk to British, European, and broader global non-American digital systems. In Britain and Europe especially, the repeated antagonistic responses of US corporations to British-European GDPR regulation, especially in the age of AI models, has led to increasing efforts by the EU to develop Free and Open Source Software (FOSS) alternatives for member-states to the software built by US private enterprise.

    In the name of making knowledge and data more readily accessible, academics could follow the lead of several local governments and EU member-states and encourage local and open source development of alternatives to proprietary software and platforms. Simple ways to do this are building and publishing datasets in ‘platform agnostic’ formats such as CSV, building surveys (and survey platforms) in Python and R, and sharing relevant (and safe) scripts to enable other academics to conduct research using the tools we have made. For better data security, it is also vital to host the software to conduct surveys and quantitative analysis locally on servers physically within the university, building literacy in basic server administration where necessary, rather than rely on vague ‘cloud’ infrastructure.

    In this paper, I will argue for a new computational ‘best practice’ focused on FOSS and on locally hosted software and data storage. I will examine the histories of platforms built and run by Microsoft, Amazon, and Google and argue that their jurisdictional position and past conduct mean that they can no longer can be relied on to fulfil professional, ethical, and legal obligations associated with research involving human subjects. I will demonstrate alternative tools and methodologies that reduce our reliance on these platforms, and argue that people with all levels of technical literacy can reliably engage with FOSS. Despite the increasing approachability of FOSS, I will argue that it is necessary that as academics we strive to improve both our individual technical literacies, and those of our peers and students.

  2. When Metadata Harmonisation Changes What We Know: Provenance, Uncertainty and the Evidential Reliability of Cultural Heritage Records in Europeana

    Qinlei Wang, King‘s College London, United Kingdom · Paper ID 202

    Abstract

    Europeana brings museum, library and archival collections into a shared discovery environment, yet the translation of institutional catalogues into interoperable descriptions raises a consequential issue for historical interpretation. A qualified attribution may appear alongside a standardised creator identifier, while an approximate date or disputed place of origin may be expressed through fields whose original evidential qualifications are difficult to recover. Such transformations affect how users understand what an institution knows, infers or continues to contest about an object.

    This proposed study examines the preservation of evidential uncertainty across Europeana Data Model mappings and semantic enrichment. Research on cultural heritage metadata quality has established the importance of completeness, multilingual accessibility and interoperability. These measures reveal considerable variation in descriptive coverage, although a technically well-formed record offers limited assurance that an attribution remains faithful to the judgement recorded by its contributing institution. Scholarship on provenance and paradata further establishes the importance of documenting how cultural heritage knowledge is produced. The investigation brings these concerns together at the point where institutional descriptions become aggregated and enriched records available to a wider public.

    Using publicly accessible Europeana records, original catalogue pages and documented mapping practices, the study will undertake a comparative audit of selected objects whose descriptions contain contested authorship, qualified dating or uncertain geographical attribution. Selection will prioritise cases in which both the institutional source and the aggregated record remain accessible, with inaccessible or changed source pages documented as limits on interpretation. Comparison will follow the wording of original qualifications into the harmonised record, examine whether contextual identifiers are accompanied by information about their provenance, and assess whether a reader can distinguish a supplying institution's assertion from an enrichment introduced downstream. Particular attention will be paid to instances where a precise identifier coexists with qualified descriptive language, since apparent precision and historical certainty represent different properties of a record.

    The analysis treats discrepancies as documentary evidence requiring interpretation, including cases where an aggregation faithfully preserves uncertainty, makes it accessible through linked source material or presents it in a less prominent form. It will avoid inferring systematic metadata loss from isolated examples and will distinguish documented mapping changes from interface-level display choices. Findings will be organised around contrasting record histories and transparently described inclusion decisions, permitting readers to examine how each interpretation follows from accessible source material. This approach also allows differences between institutional cataloguing conventions to be considered when assessing the apparent consistency of aggregated metadata.

    The article develops the idea of evidential continuity as a complement to technical interoperability. The ability to retrieve and connect cultural heritage objects across repositories is essential, while responsible reuse also depends on retaining the provisional status and institutional origins of descriptive claims. By locating these responsibilities within identifiable mapping and enrichment decisions, the proposed audit offers heritage aggregators a practical basis for examining where semantic clarity may outpace the evidence supporting it. Its contribution lies in assessing the fidelity of a record's historical qualifications as part of trustworthy knowledge integration.

  3. Documenting Craft, Transmitting Knowledge? Social Media and the Safeguarding of Hmong/Miao Intangible Cultural Heritage

    Hairong Wang, The University of Edinburgh, Scotland, United Kingdom · Paper ID 212

    Abstract

    Social media provides increasingly accessible records of intangible cultural heritage (ICH), particularly craft practices that can be photographed and filmed. Such records can make techniques visible to audiences who may never encounter them in person. However, the visibility of making does not establish what craft knowledge has been preserved, or whether it can be transmitted through a digital record. This paper examines the opportunities and limitations of social media documentation for the safeguarding of Hmong/Miao embroidery, batik and silversmithing.

    The study draws on netnography and reflexive thematic analysis of social media content relating to Hmong/Miao craft practices in China and diasporic communities. It examines posts on Weixin (WeChat), Xiaohongshu (RedNote), Instagram and Facebook, paying particular attention to demonstrations of making, accounts of learning, and responses from platform users. The analysis asks how craft processes are represented online and which dimensions of knowledge become difficult to communicate when practices are recorded and circulated as platform content.

    The analysis suggests that digital records can serve several purposes. Images and short videos preserve visible details of materials, tools and stages of production. Practitioners' accounts can give audiences access to explanations of their work, while posts about workshops can connect interest generated online with opportunities for learning in person. In these cases, documentation may support transmission by directing attention beyond the recorded content itself.

    At the same time, the apparent completeness of a visual record can be misleading. Edited demonstrations may present complex techniques as brief, repeatable sequences, with limited indication of the judgement and sustained practice required to acquire them. Posts centred on finished objects may communicate aesthetic qualities while providing little information about the community knowledge and cultural contexts that inform these practices. Such selective representations may leave unclear whose interpretations of authenticity are being presented and whose knowledge is treated as authoritative. They risk presenting fragmented or simplified accounts of craft knowledge as sufficient for understanding the practice. These limitations are particularly important for ICH as living heritage, whose transmission depends on continued practice, embodied experience and relationships between practitioners and learners, including across generations.

    The paper argues that the safeguarding value of digital craft documentation cannot be assessed through visual detail or online reach alone. Community knowledge is not merely background information: it informs how a recorded practice is understood and what may remain unspoken or invisible. Attention is therefore needed to the explanations and cultural contexts accompanying a record, as well as the opportunities for learning beyond it. For digital cultural heritage research, this calls for a more careful distinction between recording a craft practice and sustaining its transmission. The Hmong/Miao cases contribute to digital heritage research by identifying contextual and relational dimensions of craft knowledge that visual records alone may not adequately convey.

  4. Representing Cultural Context in Machine-Translated Heritage Metadata: A Document Analysis of Europeana Translate

    Sadarudeen P, University of Calicut, India · Paper ID 148

    Abstract

    Digital cultural heritage platforms bring together collections from libraries, archives, and museums, making them accessible beyond their original linguistic and geographical settings. However, this process raises several challenges beyond translation accuracy. Heritage metadata may contain references, specialised terminology, and brief descriptions shaped by historical and cultural contexts. Translating these terms without sufficient contextual understanding may lead to the loss of meaning, misinterpretation, or misleading representations of cultural heritage.

    The paper discusses the treatment of cultural context in machine-translated heritage metadata, with a focus on the Europeana Translate project. Previous work in this area has focused on the creation and evaluation of machine translation for cultural heritage collections, but this paper looks at the ways in which contextual meaning and limitations of translation are addressed in project documentation and associated guidance. The question is: How are context-sensitive translation challenges identified in the Europeana Translate material? What are the approaches presented to improve the translation of specialized heritage terminology? What other issues matter for keeping cultural meaning while improving multilingual access?

    The study will employ qualitative document analysis of publicly available project reports, technical documentation and relevant scholarly publications. The materials will be analyzed for references to contextual ambiguity, domain-specific vocabulary, multilingual vocabularies, translation evaluation and human review. The analysis will compare the challenges identified in the literature with approaches recorded by the project. It will not try to reproduce the technical evaluation of the project, nor will it claim to measure the quality of translation independently.

    The present study aims to shed light upon the relationship between the translation quality and the representation of culture in digital heritage metadata. It will examine how keeping descriptions in the original language, adding more context, and using specialist terminology resources may result in more dependable access to heritage collections. These approaches will be considered as possible practices, not as assumed solutions, because they might work differently in different languages and types of cultural material.

    By examining the documented practices and limitations of Europeana Translate, the study seeks to contribute a focused perspective on the cultural dimensions of multilingual digital heritage. It also identifies questions for future research into how translation choices influence the discovery and interpretation of cultural resources by users from different linguistic backgrounds.

  5. How Much Reality Is Necessary? Authenticity, Accessibility and Engagement in Museum Maya Indonesia, a Metaverse Museum for an Archipelago

    Eku Wand, Braunschweig University of Art (HBK), Germany · Paper ID 225

    Abstract

    Indonesia's cultural diversity – more than 1,300 ethnic groups across thousands of islands – stands in stark contrast to its museum landscape: around 450 museums, mostly underfunded and concentrated in a few cities, remain out of reach for most young people, while smartphones are almost ubiquitous. Museum Maya Indonesia (MUMAIN) addresses this gap as a persistent, multi-user online museum on a browser-based metaverse platform. With their own avatars, school classes explore tangible and intangible heritage together with their teacher, attend live lectures by scientists and curators, and follow quest-based learning paths.

    The paper unites three strands of our practice-based research in one end-to-end perspective: low-cost 3D acquisition of artefacts under difficult field conditions using a full-frame camera, a drone and iPhone LiDAR; the reduction of meshes and textures for real-time playback; and a design-thinking evaluation with 33 high-school students in a rural region, followed by a prototype test with 12 students on mobile phones, laptops and VR headsets. Laptop users responded very positively and VR users reported a markedly more emotional experience, whereas mobile users struggled to access the museum – a finding rooted in decisions taken during capture and data reduction.

    Our central argument is that authenticity in a metaverse museum is not an absolute value but a curatorial decision negotiated along the entire pipeline: How much reality does the staged space need, and how much the individual exhibit, for authenticity, vividness and didactic purpose to remain in balance? We propose a framework of trade-offs between visual fidelity, bandwidth, device accessibility and emotional engagement, complemented by interactive storytelling devices: multilingual audio guides for children who cannot yet read, AI-supported non-player characters as guides, gamified quests and downloadable instructions for traditional games.

    Finally, we discuss the societal and ethical implications of transferring heritage messages to a diverse user community: strengthening cultural identity in the spirit of "Unity in Diversity", decolonising history education in line with Indonesia's Kurikulum Merdeka, giving marginalised groups a voice, and avoiding superficial appropriation. The virtual museum complements physical institutions rather than replacing them. As an outlook, we outline the planned expansion into the culture of Nias: while the local Museum Pusaka Nias is poorly funded, more than 2,000 Nias artefacts are held in collections in Cologne, Wuppertal, Amsterdam and Leiden, whose digital capture could virtually reunite them with their communities of origin. We conclude with guidelines for metaverse heritage projects in regions with limited infrastructure.

  6. Distinctive functions of immersive projects representing historical and cultural heritage: The Ukrainian experience

    Yuliia Polovynchak, V. I. Vernadskyi National Library of Ukraine, Ukraine · Paper ID 206

    Abstract

    Ukrainian immersive projects are being implemented amid a full-scale war, constant threats of destruction to cultural heritage, and the challenges of forced migration. The research problem lies in the insufficient systematization of the functions performed by such projects in crisis contexts.

    The study aims to identify and systematize these functions based on Ukrainian cases and to determine the conditions under which the relevant experience can be applied in other societies experiencing war, population displacement, or the loss of cultural resources.

    The methodological framework comprises monitoring of the Ukrainian cultural landscape, case analysis, comparison, systematization, and generalization.

    The material analyzed consists of publicly available project descriptions, accessible virtual exhibitions, interactive reconstructions, and accompanying materials provided by their developers. The unit of analysis is an individual project, examined according to the following criteria: the object of representation, technological format, stated purpose, target audience, modes of access, and functional purpose.

    Results & Key Findings. A systematization of interrelated functional areas is proposed, as outlined below.

    The compensatory function encompasses remote access to cultural objects that have become inaccessible, including those in occupied territories, such as Crimea, or those that have been lost, looted, or destroyed in areas of hostilities, as in the case of the cultural heritage of the Kherson region. Virtual restitution and access to lost heritage.

    A separate area within this function is artistic reflection, involving art projects rather than musealization, such as those related to Chornobyl or Crimea.

    The memorialization and affective function combines the documentation of wartime destruction as an act of memorialization with the formation of public memory. Examples include the War Up Close project and VR recordings of destroyed cities.

    Closely related to the preceding function is the archival function, which involves the creation of highly accurate digital twins of cultural heritage objects. It develops into the reconstructive function, which entails the accumulation of digital documentation suitable for studying the condition of objects and supporting future restoration.

    The social-integrative and inclusive function, alongside standard measures to ensure barrier-free access and rehabilitation, entails ensuring accessibility and inclusion under conditions of martial law. In the Ukrainian context, particular emphasis should be placed on accessibility for dispersed communities comprising millions of Ukrainians who, as a result of the war, have found themselves abroad or outside their home regions.

    The decolonization and agency function consists in the representation of Ukraine’s own narrative. Ukrainian cultural heritage has long been subjected to expropriation, distortion, or appropriation within imperial and Soviet discourse.

    The scientific novelty lies in structuring the functions of immersive projects related to Ukrainian heritage according to the objectives of access, documentation, memory, integration, representation, and education, while distinguishing between realized capabilities and expected effects.

  7. What Gets Lost in Digitization? Representing Traditional Sports and Games as Living and Embodied Cultural Heritage in Digital Environments

    Mustafa Kandil, University of Pittsburgh, USA · Paper ID 55

    Abstract

    Traditional sports and games (TSG) are forms of intangible cultural heritage that are practiced, learned, adapted, and transmitted within communities. They involve not only rules and techniques, but also bodily knowledge, social interaction, cultural meanings, local variations, material practices, and relationships between practitioners and communities. Increasingly, digital technologies are being used to document, represent, teach, and communicate these traditions through digital inventories, websites, audiovisual materials, mobile applications, digital games, and online educational resources. While these initiatives can increase documentation and access, an important question remains: how well can digital environments represent heritage that is fundamentally living, embodied, and socially practiced?

    This paper examines what happens when traditional sports and games are translated into digital forms. It asks: What dimensions of traditional sports and games as living and embodied heritage are captured, transformed, or difficult to represent through digital technologies?

    The study uses a comparative qualitative analysis of selected digital initiatives related to traditional sports and games. Cases will represent different forms of digital engagement, such as heritage inventories, audiovisual documentation, interactive or playable environments, and educational resources. The analysis examines how these initiatives represent key dimensions of TSG, including rules and techniques, bodily practice, cultural context, local variation, practitioner knowledge, social interaction, learning, and transmission.

    The study approaches digitization not simply as a technical process of recording heritage, but as a process of selection and representation. Digital environments may effectively document rules, terminology, visual information, recorded performances, and certain forms of contextual knowledge. However, dimensions such as tacit knowledge, bodily experience, face-to-face learning, community relationships, local variation, and the social settings in which traditional games acquire meaning may be more difficult to represent digitally.

    Through comparison across different types of digital initiatives, this paper aims to identify both the possibilities and limitations of digitally representing traditional sports and games. Rather than arguing that digital representation either preserves or diminishes living heritage, the study investigates which dimensions are represented effectively, which are transformed, and which remain difficult to capture. The findings seek to contribute to more holistic approaches to digital cultural heritage by demonstrating the importance of considering embodied practice, cultural context, and community knowledge alongside technical documentation and digital access.

  8. Visual storytelling and project-based learning in the preservation of cultural heritage: the experience of Kyiv National University of Culture and Arts

    Maryna Tolmach, Kyiv National University of Culture and Arts, Ukraine · Paper ID 227

    Abstract

    This paper explores the incorporation of cultural heritage into project-based design education at Kyiv National University of Culture and Arts. It emphasises visual storytelling as a means of interpreting historical materials, addressing contemporary issues, and engaging audiences beyond the university. The study examines how educational projects blend research, artistic production, and digital documentation in response to the cultural impacts of Russia's war against Ukraine.

    The study applies a qualitative approach based on an analysis of specific projects implemented at the Kyiv National University of Culture and Arts: ‘Creative Resistance’, ‘Heritage at Risk’ and ‘Wonders of Trypillia’. The analysis is based on project descriptions, visual works, digital collections and exhibition documentation. It examines four aspects: the use of historical and contemporary sources; the transformation of research findings into visual narratives; the development of students’ interdisciplinary skills; and the formats through which the audience is introduced to and explores the projects.

    “Creative Resistance” involves students, educators, and graduates in producing posters that respond to the war. Its catalogue-calendar, “365: Chronology of Creative Resistance”, comprises 365 works arranged as a visual account of the first year of the full-scale invasion. Their presentation through exhibitions, public activities, and a digital collection allows the paper to examine the relationship between contemporary graphic art, public communication, and the documentation of wartime experience.

    “Heritage at Risk”, developed within the same creative community, uses paired posters to relate Ukrainian cultural traditions to wartime destruction. One poster presents a tradition or artistic practice; the other employs its motifs and artefacts to address a specific humanitarian or environmental catastrophe. QR codes provide access to additional contextual information. The exhibition format, including its presentation in Zagreb, offers a case for examining how visual juxtaposition and supplementary digital content support the interpretation of unfamiliar cultural material.

    “The Wonders of Trypillya” concerns the interpretation of archaeological heritage through digital modelling and collection development. Students study museum objects and produce 3D models of ceramic reconstructions by craft artist Liudmyla Smoliakova based on archaeological evidence. The case raises questions about the documentation and presentation of relationships between archaeological sources, contemporary reconstructions, and their digital representations.

    Research constitutes a required stage of project development, informing the selection of material, its interpretation, and its presentation. The projects combine graphic design with digitisation, 3D modelling, metadata preparation, and digital collection management. Cooperation with cultural institutions, practitioners, and exhibition partners situates educational assignments within professional practice and public concerns.

    The paper identifies shared principles of project organisation and differences in the narrative and presentation strategies used across the cases. It contributes a practice-based account of heritage-oriented design education, with particular attention to the relationship between research, contemporary artistic interpretation, and opportunities for audience engagement.

  9. Analysing Interpretive Authority in Games to Inform Digital Cultural Heritage Design

    Sha Huang, The University of Edinburgh, Scotland, United Kingdom · Paper ID 216

    Abstract

    Digital cultural heritage experiences increasingly use interactive narratives and game mechanics to engage audiences with the past. Yet making interpretation playable raises questions beyond knowledge acquisition concerning the authority, uncertainty, and revisability of cultural meanings. While puzzles often rely on identifiable solutions and clear feedback, cultural heritage interpretation may involve incomplete evidence, situated knowledge, and multiple accounts. This tension presents a design challenge for games intended to support thoughtful engagement with cultural materials.

    This paper examines that challenge through a qualitative comparative analysis of Heaven's Vault (inkle, 2019), Chants of Sennaar (Rundisc, 2023), and Pentiment (Obsidian Entertainment, 2022). The games are selected as contrasting models of interpretive interaction, rather than direct representations of specific real-world heritage communities. Heaven's Vault positions translation as provisional archaeological inference; Chants of Sennaar links contextual decipherment to explicit validation and cross-cultural communication; Pentiment requires players to make historically situated judgments despite incomplete evidence and uncertain outcomes.

    The study uses structured close reading of representative gameplay sequences to examine the encounter, interpretation, recording, testing, and practical use of evidence. It develops four comparative dimensions: the visibility and attribution of interpretive sources; the range of hypotheses available to players; the mechanisms through which interpretations gain confirmation or closure; and opportunities for revision alongside the consequences of adopting a particular account. Through these dimensions, the paper investigates provisional inference, system-validated decoding, and consequential judgment as differing configurations of interpretive authority, while attending to their respective affordances and limitations.

    The comparison then informs a prospective design application within an ongoing museum interpretation project at National Museums Scotland (NMS), focusing on Hmong/Miao textile heritage. Hmong/Miao refers here to culturally diverse ethnic communities rooted in southwest China, with related communities across Southeast Asia and the diaspora. The application explores a possible interaction in which players examine a textile object, propose provisional interpretations of its motifs, encounter contextual information and accounts from different sources, and reconsider their initial assumptions. Rather than requiring one definitive answer, the proposed interaction would make interpretive sources, uncertainty, and revisions visible. The application remains an illustrative design scenario, not an implemented or empirically evaluated intervention.

    The paper contributes a focused analytical vocabulary for comparing interpretive gameplay and a set of provisional design principles for digital heritage experiences. It considers the capacity of interfaces and rules to support inquiry without prematurely fixing cultural meaning, while keeping interpretive agency attentive to evidence, attribution, and accountability. Future iterative development and community-informed evaluation are identified as necessary steps for assessing these propositions in practice.

  10. Digital Twin Technologies for Cultural Heritage Sites: A Conceptual Framework for Documentation, Monitoring, and Conservation

    Jülide Edirne, Haliç University, Türkiye · Paper ID 134

    Abstract

    The preservation and sustainable management of cultural heritage sites have become increasingly challenging due to environmental degradation, climate change, urbanization, natural hazards, and human-induced pressures. Traditional documentation and conservation approaches often rely on fragmented datasets, periodic inspections, and independent monitoring systems, limiting their capacity to support integrated and proactive heritage management. In recent years, Digital Twin (DT) technologies have emerged as promising tools for integrating physical environments with dynamic digital representations, enabling enhanced documentation, monitoring, analysis, and decision-making. However, their application to cultural heritage remains challenging due to the diversity of heritage assets, the complexity of historical environments, and the lack of standardized frameworks for integrating heterogeneous data sources.

    This study aims to develop a conceptual Digital Twin framework for the documentation, monitoring, and conservation of cultural heritage sites, extending beyond individual historic buildings to encompass historic urban environments, archaeological sites, and cultural landscapes. The proposed research adopts a structured literature review and comparative analytical approach to examine existing Digital Twin applications in cultural heritage management. Particular attention is given to the integration of Heritage Building Information Modeling (HBIM), Geographic Information Systems (GIS), three-dimensional documentation technologies, Internet of Things (IoT) sensors, and environmental and structural monitoring systems.

    Based on the identified technological capabilities and research gaps, the study proposes a five-layer conceptual framework comprising: (1) physical heritage assets and their surrounding environments, (2) digital representation and spatial documentation, (3) data acquisition and interoperability, (4) monitoring and analytical processes, and (5) conservation-oriented decision support. The framework is designed to accommodate different spatial scales, data availability levels, and conservation requirements while facilitating connections between physical heritage environments and their evolving digital representations.

    To examine the practical applicability of the proposed framework, an illustrative implementation scenario involving a cultural heritage area is considered. This scenario focuses on the integration of spatial documentation, environmental observations, structural condition indicators, and conservation-related information within a unified digital environment. Particular emphasis is placed on data interoperability, scalability, and the potential transition from static digital documentation toward continuously updated Digital Twin systems. The proposed approach also distinguishes between digital models, digital shadows, and fully connected Digital Twins, recognizing that real-time data exchange and feedback mechanisms represent essential aspects of advanced Digital Twin implementation.

    The anticipated contribution of this research is a transferable conceptual framework that supports a systematic understanding of Digital Twin technologies across different categories of cultural heritage sites. By connecting documentation processes with monitoring strategies and conservation decision-making, the study seeks to provide a methodological foundation for future research and practical applications. Furthermore, the proposed framework highlights opportunities for interdisciplinary collaboration among architecture, civil engineering, heritage conservation, and geospatial information sciences. The study is expected to contribute to ongoing discussions on digital transformation in cultural heritage management and to support the development of more integrated, adaptable, and sustainable conservation strategies.

  11. 09:30–09:40

    Coffee Break

  12. 09:40–12:00

    Session 7: Beyond Digitisation: Unlocking the Power of Cultural Heritage - Use & Reuse - Part II

  13. Panagia Old Town of Kavala - Conceptual Model

    Rafael Prezado, University of Évora, CIDEHUS, Portugal · Paper ID 137

    Abstract

    In the context of the INT-ACT (Intangible Cultural Heritage, Bridging the Past, Present, and Future) project, the tangible and intangible cultural heritage of the historic old town of Panagia, in Kavala, is studied. Kavala constitutes one of the most layered urban heritage sites of the Ottoman period in northern Greece, encompassing a fortified acropolis, a monumental aqueduct, religious architecture, and elite residential buildings, each bearing the material traces of successive construction phases and transformations. An ontological model was developed and instantiated with four of its sites, namely the Kamares Aqueduct, the Kavala Fortress, the Halil Bey Mosque, and the Mehmet Ali Pasha House.

    Each site is divided into morphological building sections, distinguishing solid structural elements from voids, while its stratigraphic building units capture successive constructive phases through events of part addition and part removal. This allows each building to be represented in its temporal biography and trace transformations, for example the Ottoman reconstruction of the fortress following the destruction of the earlier Byzantine acropolis. The model is aligned with CIDOC-CRM and CRMba, and proposes new concepts which specialize several aspects for documenting the studied buildings.

    This model is related to a Demonstrator, which seeks to counter the risk of intangible cultural heritage fading from collective memory by developing methods to extract, structure, and present its emotional, experiential, and environmental dimensions as formalized, machine-readable knowledge. As such, the Demonstrator addresses Kavala as a designated heritage site, exploring how a community museum, built around the old town's monuments, can promote community engagement around the preservation of both tangible and intangible heritage. A curated body of narrative content creates a multisensory XR environment combining auditory and visual dimensions.

  14. From Disclosure to Return: Designing Accountable AI-Assisted Interpretive Pathways for Cultural Heritage

    Andrew Ahachinsky, Bobbinet, Canada · Paper ID 63

    Abstract

    AI-enabled heritage interfaces increasingly offer summaries, thematic pathways, translation, search assistance and collection narratives at a scale institutions could not previously sustain. Their apparent fluency, however, can conceal the relation between a public-facing claim and the record, catalogue metadata, rights conditions, community knowledge and institutional judgment that make interpretation accountable. This conceptual paper proposes “disclosure to return” as a design and governance framework for AI-assisted cultural-heritage interpretation.

    The framework has four linked practices. First, provenance disclosure identifies whether an output is grounded in a specific record, catalogue metadata, a rights-cleared interpretive source, a composite dataset or synthetic generation. Second, contextual accountability makes uncertainty, omissions, contested descriptions and responsible institutional actors legible where users encounter the output, rather than burying them in technical documentation. Third, reciprocal access provides practical routes for users, record creators and affected communities to question, correct, add context or decline a framing. Fourth, return records how those interventions subsequently change descriptions, interfaces, training data or governance.

    Rather than claiming an empirical study, the paper offers a practice-oriented conceptual model developed through critical synthesis of archival stewardship, participatory interpretation, information ethics and responsible-AI scholarship. It asks how digital-heritage systems can make their interpretive conditions inspectable without reducing public access to a compliance exercise. The paper connects provenance and explainability to human-centred user experience: a visitor should be able to ask what an output is based on, what it does not establish, who is accountable, and what can happen when they respond.

    The contribution is a compact evaluative vocabulary for cultural-heritage practitioners designing AI-mediated access. It reframes AI not simply as an automation layer or a neutral retrieval mechanism, but as an institutional relationship whose claims must remain contestable, revisable and answerable to the records and publics it mediates.

  15. From AI-Generated Heritage Narratives to Evidence-Based Interpretation: A Design Framework for Critical Digital Heritage Education

    Angelo La Licata, University of Sassari, Department of History, Human Sciences and Education, Italy · Paper ID 136

    Abstract

    Generative AI can transform museum records into fluent heritage narratives while obscuring the difference between a documented feature, an attributed interpretation and a later intervention. This paper asks how an educational narrative can preserve those differences and make their justification available to learners. It proposes a design framework that connects each consequential assertion to a particular source passage, preserves its qualification and identifies the historical layer to which it refers. The framework adapts existing approaches to factuality and heritage documentation into a procedure for interpreting and revising claims that preserves attribution and leaves institutional interpretations open to further inquiry.

    The documentary design study examines three Egyptian objects in The Metropolitan Museum of Art's digital collection: the faience hippopotamus known as William (17.9.1), a granary model with scribes (20.3.11), and a kneeling statue of Hatshepsut (30.3.1). Publicly accessible catalogue descriptions and metadata are supplemented by William's published audio-guide transcript and a museum conservation account. Twenty-one purposively selected audit entries are nested within these three records and their supporting documents. One axis distinguishes recorded statements, interpretations, unresolved questions and unsupported propositions; the other locates ancient production or use, excavation and collection, conservation, and modern reception. Source passage, communication channel, qualifier and editorial decision are recorded alongside them.

    A documented difference emerges within William's institutional record: the catalogue describes deliberate ancient breakage as probable, whereas the published audio-guide transcript presents the explanation declaratively. The analysis concerns the transcript, without attributing an intention to that difference or treating the channels as independent corroboration. The granary case shows why counts within a funerary representation cannot directly establish historical workforce ratios, while the suggested protective function of its raised corners remains tentative. Hatshepsut's record requires separation of the museum's inscription reading, proposed placement and modern recomposition; the relationship between its find-context fields and descriptive chronology remains unresolved within the corpus. Worked revisions of deliberately constructed sentences demonstrate these distinctions without presenting those sentences as captured model outputs.

    The proposed teaching activity asks advanced secondary or introductory university learners to revise a short object narrative and justify their decisions through the claim register. It makes source location, qualification, historical layers and the suspension of unsupported conclusions available for explicit assessment. These are proposed descriptive criteria rather than a validated scale. The contribution is a traceable documentary demonstration and a specified teaching design, grounded in actual published materials. The entries do not constitute independent observations from which model error rates could be calculated, and no inter-rater agreement or learning outcomes were measured. The cases show how narrative continuity can be retained while leaving the reasoning behind an interpretation open to examination.

  16. Taime Walker: Learning Neolithic Skills with a Virtual Reality Game

    Jonathan Barbara, Saint Martin's Institute of Higher Education, Malta · Paper ID 142

    Abstract

    Serious games are an area of strong interest for research on the valorisation of cultural heritage. This thesis project explores the pedagogical effectiveness of an approach based on structured missions within an exploratory virtual reality environment. The aim is to foster the acquisition of procedural knowledge, making the learning process more effective and enjoyable.

    The project involved a design and technical implementation phase using a game engine and a Virtual Reality headset, focusing on the creation of interactive objects and intuitive game mechanics. The experimental validation analysed users’ engagement and learning in two different interaction scenarios: the first was guided by missions, while the second was geared towards free exploration of the environment.

    The results suggest that the introduction of guided missions may foster a deeper understanding of the presented scenario. However, the validation is not considered sufficient to have scientific relevance, owing to its limited scope in terms of sample size and heterogeneity of the test participants.

  17. Bridging Monitoring and Interpretation: HeriTwinneD, a Web Application for Digital Twins of Heritage Buildings

    Rana Tootoonchi, University of Bamberg, Germany · Paper ID 126

    Abstract

    Digital twins are increasingly being explored as tools for the conservation of cultural heritage. However, the capabilities required for a heritage digital twin are often distributed across separate systems. Environmental monitoring platforms can display continuous data and generate risk indicators but often provide limited spatial context. By contrast, building viewers and HBIM platforms can present detailed 3D models and historical information, yet are rarely connected to live environmental data and interpretation. Integrated solutions remain uncommon and are often restricted to commercial platforms, not open source, or built for a single project. Users therefore benefit from either environmental data without a meaningful building context or detailed models that do not reflect the monitored condition of the building.

    HeriTwinneD addresses this gap through a web application that brings together interactive 3D models, environmental monitoring, historical documentation, accessible interpretation, and conservation guidance. The project is being developed within the Bamberg Graduate School for Smart City Science. The study begins with a focused review of heritage dashboards and digital twins, examining how existing approaches connect the main components of a heritage digital twin. The review finds that the main challenge is not the absence of individual technical components, but their limited integration and interpretation in heritage applications.

    The web app uses ten Efento NB-IoT loggers recording temperature and relative humidity every fifteen minutes in two historic buildings within Bamberg World Heritage Site (The Old Town Hall and the Former Dominican Church). Each logger is assigned to its position in the 3D model, where its current status is shown as a colored marker and its related measured data over time is transferred to Efento cloud and accessible on web app through REST API. The web application is built using HTML, CSS, JavaScript, and the Angular framework, while Babylon.js supports the 3D visualization.

    During Open Monument Day in Bamberg in September 2026, the web app was initially evaluated based on 30 questionnaires from public visitors and heritage professionals. Participants particularly valued a clear 3D model of the building, detailed information on rooms and architectural features, and explanations of how environmental conditions affect heritage buildings. Raw monitoring values were less prioritized, suggesting that data should be interpreted to be more meaningful for the public users. A color change on the model was the preferred alert mechanism. Respondents also wanted to understand how the sensors and models were developed and access the application on site through QR code. These findings support a layered interface that combines accessible visual interpretation for general users with more detailed monitoring information for professionals. The findings will later be used to refine and develop the web app.

    This study combines a review of current heritage digital twins with the development and initial user evaluation of a working web application. It shows that a useful heritage digital twin should do more than bring together 3D models, sensor data, and documentation. It should help both visitors and professionals understand what changing environmental conditions mean for a building and support practical conservation decisions and public engagement.

  18. AI for Knowledge Extraction in Heritage Collections: Enriching 3D Assets with Explainable, Vocab-Aware Semantic Annotation

    Cordula Hansen, XYZ Technical Art Services GmbH, Germany · Paper ID 131

    Abstract

    A central challenge in digital cultural heritage is moving from capturing objects to presenting 3D assets as part of a knowledge system by attaching knowledge to them. Knowledge extraction, the automated process of transforming raw data into machine-readable, interpretable knowledge representations suitable for automated reasoning has elevated the capabilities of machine learning from simply describing a 3D asset. It can propose meanings, relationships and contexts that make collections searchable, comparable and reusable. Yet for such machine-generated enrichment to be trusted in scholarly and public settings, it must be more than a label: it must be explainable, vocab-anchored, auditable, and embedded in a knowledge system that connects the object to the people and values it carries.

    This paper examines how machine learning can be used for knowledge extraction in 3D heritage collections, and how the results of that extraction can remain trustworthy and reusable. We present Culturama's AI-powered, EDM-compliant enrichment as a case of vocab-anchored, auditable semantic annotation: model-generated annotations are suggestions presented to users for review and grounded in controlled vocabularies - Getty AAT, GeoNames and VIAF - and harmonised across 24 EU languages, so that each assertion is traceable to a defensible reference rather than to an opaque model output. This enrichment sits within an automated harmonisation pipeline that converts and aligns heterogeneous sources such as meshes, point clouds and Gaussian splats, as well as metadata formats, into web-optimised geometry derivatives and a common metadata schema (EDM), making otherwise incompatible 3D and metadata assets jointly searchable and reusable.

    Beyond a systems description, we make a conceptual argument. We contend that explainable, provenance-documented annotation is what makes machine-generated heritage knowledge reusable at all: The trustworthiness of an AI annotation depends on the ability to trace it to its source asset, its grounded vocabulary term, and the model and confidence context in which it was produced. Documenting this provenance is therefore not a feature but an architectural requirement, and it is for this reason that Culturama builds it into the core of the platform.

    Building on this, we offer a working definition and implementation of the Memory Twin, the successor concept to the digital twin, which ties a digital representation to meaning, value and community. We show how Culturama's knowledge layer feeds a template-based experience builder that enables collaboration between cultural heritage institutions (CHIs) and the cultural and creative industries (CCIs) to turn assets into interactive, embeddable, audience-facing stories. In doing so, the platform connects the stewardship of heritage with its creative reuse, bridging research and engagement, and keeping European cultural heritage accessible, available and protected against appropriation and misuse.

    The paper's contribution is thus twofold. First, it provides a theoretical model of dataset harmonisation and vocab-aware semantic annotation for 3D heritage, articulating the conditions under which model-generated enrichment is trustworthy for scholarly use. Second, it demonstrates this model in a working, production implementation that unifies AI enrichment, provenance, harmonisation and public-facing storytelling within a single knowledge system.

    We close with a discussion of limitations, in particular the tension between explainability and usability for diverse audiences, and a research agenda for standardising the provenance of machine-generated heritage knowledge across institutions and industries.

  19. A Concept for a Community-Managed Interactive Ancient Map for Information Access and Preservation of the Ciela Map

    Yoki Purwadi, Institut Seni Budaya Indonesia (ISBI) Bandung, Indonesian · Paper ID 117

    Abstract

    Ancient maps record geographical information, place names, and depictions of the space and daily life of past communities. However, their information is not always easy to read and trace, because they use scripts, symbols, and visual representation systems that differ from modern cartographic conventions. This problem is evident in the Ciela Map, a sixteenth-century continental map depicting the Tatar Sunda region, particularly Timbanganten, from a local perspective. The original has been lost; what remains in Ciela Village, Bayongbong, Garut Regency, is a copy on cotton cloth made by K.F. Holle in 1862. The map records place names, rivers, mountains, and other geographical features written in a local script, drawn with distinctive graphic symbols, and marked with letters and numbers that require explanation before they can be identified. Beyond legibility, its existence as a physical artefact raises problems of access. Because the map also serves as an object of ritual, cultural tourism, education, and research, requests to view it require the artefact to be opened, unfolded, and handled repeatedly, increasing the risk of physical damage. An alternative way of presenting its information is needed, one that widens access while supporting preservation of the original.

    This study formulates the concept of an interactive ancient map that helps users read, interpret, and trace information in the Ciela Map while reducing the need for direct access to the artefact. It uses a descriptive-analytical approach with a conceptual orientation. The stages comprise identifying the map's physical characteristics and information through visual analysis of photographic documentation of the artefact and its facsimile, mapping access and information-retrieval needs, and formulating the design's components and functions. The formulation is also informed by previous studies of the Ciela Map.

    The concept includes high-resolution digital images, magnification of map sections, marking of features such as rivers, mountains, and place names, and supporting information on place names, symbols, and geographic context. Content management is designed so that the owners or the community keeping the map can update and present information without heavy reliance on IT specialists.

    Unlike digital map presentations that focus on displaying images and are managed by well-resourced institutions, this concept integrates three interests in a single design: legibility of information, tracing of map features, and reduced handling of the original artefact. It is designed to be managed independently by the owners or the community keeping the map. Interactivity is understood not merely as an image that can be enlarged, but as an approach that links the visual representation of the map, including non-conventional scripts and symbols, with contextual information and preservation needs.

    The outcome is a concept and functional design that serves as a reference for developing an interactive ancient map application, not a product that has been implemented and tested. The approach could be adapted to other ancient maps managed by their owners or communities.

  20. What Wartime Ukraine Reveals About the Boundaries of Digital Cultural Heritage

    Daniil Shmatkov, Scientific and Research Institute of Providing Legal Framework for the Innovative Development of National Academy of Legal Sciences of Ukraine, Ukraine · Paper ID 109

    Abstract

    The war in Ukraine has demonstrated the importance of preserving cultural heritage in forms that can endure the loss or displacement of physical objects. Museum professionals, archivists, conservators and volunteers have worked to document damaged sites, secure collections and maintain access to cultural records under exceptional conditions. Their efforts coincide with declining museum revenues, rising preservation costs and the loss of established funding sources. Digitisation has consequently become essential to emergency safeguarding, public access and the search for new sources of revenue. The Ukrainian experience brings into focus challenges relevant to cultural institutions well beyond conflict zones. The study draws on Ukrainian initiatives involving 3D documentation of historic buildings damaged by Russian attacks and street art created amid the ruins, as well as virtual reconstructions of destroyed theatres and other architectural landmarks using photographs, plans and video recordings. It also examines physical and virtual exhibitions documenting artworks looted during occupation, emergency archiving of endangered cultural collections, NFT fundraising for museums facing financial hardship, online collections of children’s drawings depicting the invasion, and the commercial use of museum reproductions. The research combines examination of publicly accessible collections, project records, technical documentation, institutional guidance, platform policies and judicial decisions with an analysis of Ukrainian law in force in 2026 and relevant EU directives. Selected national decisions are considered where they clarify the application of European rules. The analysis considers the relationship between the methods used to produce digital representations and the permissions needed to preserve, exhibit, adapt or commercialise them. It also examines the responsibilities that arise when cultural institutions work with independent developers, international archives and technology platforms. The legal status of a digitised cultural object depends on its original form, the materials used and the methods of reproduction. A museum may own a historical artefact without holding rights in the photographs used for its reconstruction, while the resulting model may qualify for separate protection. Permission to create a preservation copy may also differ from that required for online publication or commercial reuse. Cross-border archives complicate the management of permissions for materials from multiple sources, while dependence on external platforms can undermine long-term access to cultural collections. The findings identify the areas in which existing rules support these practices, the uncertainties that emerge in their application and the challenges that remain under Ukrainian and EU law. By examining these questions in the context of building and maintaining digital collections, the study connects legal analysis with cultural heritage information management, including provenance, responsible reuse and long-term accessibility.

  21. Conversational artificial intelligence for Archaeological Heritage: interpretative challenges and public engagement in medieval museums

    Stefano Bertoldi, Università di Siena, Italy · Paper ID 101

    Abstract

    Conversational artificial intelligence offers possibilities for communicating archaeological knowledge, but its application to historical characters raises questions concerning evidence, interpretation, and the authority of generated narratives. This paper examines Johannes, a voice-based digital character installed at the Museum of Miranduolo (Tuscany, Italy), developed from the archaeological reconstruction of a twelfth-century individual associated with the medieval settlement.

    Rather than presenting an omniscient virtual guide, the project constructs a historically situated interlocutor whose knowledge is constrained by chronology, geography, social position, and archaeological and documentary evidence. Its architecture combines a conversational language model with curated knowledge, a reconstructed cognitive geography, and a supervised memory system. Proposed additions to the character's memory undergo scholarly review, preventing generated narratives from automatically acquiring the status of historical evidence.

    The paper also discusses an analysis of museum interactions recorded between July and September 2026. A corpus of 1,016 extracted narrative fragments reveals recurring interests in the castle, everyday life, and warfare. Because these fragments do not represent complete visitor conversations, the analysis distinguishes observed interaction patterns from unsupported claims about audience demographics or engagement.

    Particular attention is given to hallucinations, chronological intrusions, and the tendency of fluent first-person narration to obscure the boundary between evidence and conjecture. Johannes demonstrates how conversational agents can support public engagement by making archaeological interpretations accessible through dialogue, while simultaneously functioning as diagnostic tools that expose gaps and inconsistencies in historical reconstruction. The central challenge is therefore not to simulate medieval consciousness, but to develop accountable forms of artificial memory.

  22. Digital Games and Local Heritage: Sense of Belonging and Historical Identity among Portuguese Children

    Filipe Oliveira, University of Minho, CIEd, EDHILAB, Portugal · Paper ID 89

    Abstract

    Heritage has a history. It is the expression of a community and of its culture, in the specificities and convergences that this culture shows over time, and for that reason it is a factor of identity. Its educational exploration can begin in the first years of schooling, as Cooper (2004) argues. For her, children “are able to work with concepts of validity and to solve increasingly complex historical questions, and can learn abstract historical concepts by linking them to concrete ones”, which is why this work “should begin early” and should not be restricted “to the repetition of factual information”.

    This research is part of a Local Heritage Digital Archive project, which aims to apply digital technologies to heritage and to promote community involvement. It was developed in partnership between a primary school, a municipal institution for culture and science, and a public university in the North of Portugal, and involved 50 children from preschool and the first cycle of basic education. The initiative started from the (de)construction of the concept of heritage as a living laboratory of practices, knowledge and meanings that cross historical time and space, and whose dynamic of change is interrelated with intergenerational material and immaterial inheritance.

    The study aims to understand how the combination of digital games (puzzles, crosswords, memory games, association games, among others) and challenges related to local heritage can help these children to recognise, interpret and value it as a reference for identity and as a way of relating to the past, the present and the horizon of expectation. In this context, the digital resources were not designed as an end in themselves, but as guides for the relationship between the local and social memory, combined with direct contact with historical sources in different formats and spaces.

    Methodologically, this is a qualitative approach centred on an intervention in six sequential educational moments. Data were collected throughout the intervention, from the children’s productions in the various tasks and through semi-structured interviews with the participants. The analysis, inspired by Grounded Theory, was organised around two constructs, historical culture and historical identity. Once the analysis of the interviews is complete, the paper will discuss to what extent children’s historical thinking expresses a sense of belonging, and what role digital games played in historical understanding.

  23. Digital Cultural Heritage as a Tool for Adolescent Resilience: Serious Games, Social-Emotional Learning, and Inclusive Education

    Eleni Christodoulou, Ministry Of Education- Regional Educational Advisor for School Social Workers, Greece · Paper ID 146

    Abstract

    Adolescence is a critical developmental period during which identity formation, emotional regulation, social belonging, and resilience are shaped by individual, interpersonal, and cultural influences. At the same time, digital cultural heritage offers emerging opportunities to transform cultural knowledge into meaningful, interactive, and inclusive learning experiences. However, its potential contribution to adolescent psychosocial development remains an important area for interdisciplinary exploration. This paper proposes a conceptual framework that positions digital cultural heritage as a resource for promoting adolescent resilience through the integration of adolescent medicine, social work, serious games, and inclusive education.

    The proposed framework explores how immersive cultural narratives, digital storytelling, and heritage-based serious games may create opportunities for social-emotional learning, empathy, intercultural understanding, critical thinking, and a stronger sense of belonging. Drawing on the biopsychosocial perspective of adolescent medicine and the strengths-based, person-centred, and rights-based approaches of social work, the framework conceptualises cultural participation as a potential protective resource that may support psychosocial wellbeing and social inclusion. Particular attention is given to adolescents experiencing social vulnerability, educational disadvantage, or barriers to participation.

    The framework also addresses the ethical implications of technology-mediated engagement with young people, including privacy, informed consent, digital accessibility, cultural sensitivity, equitable representation, and the responsible use of artificial intelligence. It emphasises the importance of co-designing digital heritage experiences with adolescents, rather than treating them solely as recipients of educational content. Potential evaluation dimensions include user engagement, social-emotional learning, perceived belonging, inclusiveness, and psychosocial outcomes.

    By bridging digital cultural heritage research with adolescent medicine, social work, and educational practice, this paper advances a human-centred and ethically informed perspective on the social value of digital heritage. It proposes a foundation for future interdisciplinary research and participatory interventions that explore how cultural heritage technologies can contribute to more inclusive learning environments and support the developmental needs of adolescents.

  24. A Low-Cost Phygital Workflow for GLAM: Digitising Physical Dioramas and Living History into Web-Accessible Cinematic VR

    Martin Debattista, Institute of Tourism Studies, Malta · Paper ID 123

    Abstract

    A Low-Cost Phygital Workflow for GLAM: Digitising Physical Dioramas and Living History into Web-Accessible Cinematic VR.

    Small and independent galleries, libraries, archives and museums (GLAM) moved towards digital interpretation during the Covid-19 pandemic with fewer resources than larger ones. Much research on extended reality (XR) in cultural heritage assumes industrial-grade capture and specialist teams, and treats physical interpretive media as something digital technology replaces. This paper instead presents a low-cost phygital workflow that turns scale-model dioramas and historical re-enactment into cinematic virtual reality (CVR), viewable on a headset, on visitors’ own smartphones or as web VR.

    The workflow was developed through practice-as-research on a counter-factual case: the Axis invasion of Malta (Operazione C3 / Unternehmen Herkules) planned for summer 1942 but never executed. Dioramas of an Axis airfield, the Italian invasion fleet and a land battle were captured with consumer 360-degree cameras at figurine eye level, placing viewers among the troops, aircraft and ships. Re-enactments were filmed on location, including a Bofors gun crew, a coastal pillbox defence and a reconnaissance flight in a vintage WWII Piper Cub reconnaissance plane. ThingLink hosts the result as a web-based experience with hotspots.

    Visitors to a scale-model exhibition at the Malta Maritime Museum (2025) viewed the invasion-fleet diorama and then its CVR capture, and completed a questionnaire (n = 23). Ten experts (curators, historians, re-enactors, modellers and game developers) were interviewed. Of the 23 visitors, 22 preferred diorama and VR together; none chose VR alone. Expert interviewees described being “transported to the past”, especially in re-enactments filmed at authentic locations, and saw VR as complementing the physical model, not replacing it.

    The paper also reports what did not work: detail lost in 360-degree capture, unusable smartphone 3D scans, failed green-screen experiments, plat-form limits on 360-degree video in VR mode, and headset discomfort among older users. It derives guidelines for small institutions: consumer-grade capture, open-source software, delivery to personal devices and early multidisciplinary protocols. The workflow forms part of the Phygital Ex-tended Reality Heritage Interpretation (PhXR) Framework and shows that small GLAM institutions can extend physical collections into immersive experiences without a large budget.

  25. "Kurioso" - Mediating Cultural Agency: A Randomized Controlled Experiment on Generative AI in an Online Participatory Museum

    Jasmin Jalil, University of Rome La Sapienza / University of Turin, Italy · Paper ID 118

    Abstract

    Museums and cultural institutions are exploring AI to communicate heritage to wider audiences, but the point at which AI involvement compromises audiences’ experience and their ability to contribute to cultural meaning-making remains little examined empirically. Studies of human–AI co-creation have mostly used laboratory or generic creative tasks and tend to compare “AI” with “no AI”; few have examined intermediate degrees of involvement, especially in cultural settings, where agency is crucial. To address this knowledge gap, this paper presents “Kurioso”, a doctoral research project centred on a purpose-built gamified web platform that works as an online participatory museum, or “cultural social network”. Participants, recruited online as volunteers, choose artworks from open collections and curate their own exhibition, writing a title and a curatorial text. They are randomly assigned to one of four levels of AI support, from no AI to a text generated entirely by the AI, and do not know which level they received until after their first exhibition, which makes it possible to compare their experience before and after they find out. The evaluation combines what participants do (how long they take, how much they edit the AI’s text, whether they abandon the task) with what they report (their sense of control, ownership, creativity, enjoyment, perceived quality and learning), to test whether behaviour and perception converge. The platform is operational, and data collection begins in November 2026. By comparing levels of AI support on the same task, the study aims to show how experience, learning and cultural agency change as the AI takes on more of the work, and whether there is a “sweet spot” at which AI helps without displacing the user. This can give museums and cultural institutions an evidence-based way to decide how much AI to integrate into their cultural services. And because Kurioso can host other collections, languages and audiences, the same question can be examined across different institutions and cultural contexts.

  26. IAU-All Skies Encyclopaedia

    Susanne M Hoffmann, International Astronomical Union, France · Paper ID 116

    Abstract

    Since its inception in 1919, the International Astronomical Union (IAU) has been in charge of global standards, e.g. astronomical units of measure, the boundaries of constellations, and nomenclature for celestial objects. Star names are defined by the Working Group on Star Names (WGSN). For their proper documentation, the WGSN created the All Skies Encyclopaedia (ASE). https://ase.exopla.net/ It serves as an online repository for intangible cultural heritage like names of stars and constellations, their transformative history, etymology and cultural significance.

    My contributions will briefly present the ASE.

  27. 12:00–13:00

    Lunch Break

  28. 13:00–14:30

    Session 8: Beyond Digitisation: Unlocking the Power of Cultural Heritage - Use & Reuse - Part III

  29. Rethinking Photo Album Representation Through IIIF

    Julien Antoine Raemy, Walter Benjamin Kolleg, University of Bern, Switzerland · Paper ID 57

    Abstract

    This contribution investigates how semantic web infrastructures for cultural heritage can represent objects whose meaning depends on material states that conventional metadata models flatten away. We take as our case Album 31, a 1920s vernacular photo album from the Kreis Family Collection (Cultural Anthropology Switzerland, Basel), whose pages exist in multiple simultaneous configurations — with and without protective interleaves, with and without loose photographs — that standard single-image digitisation cannot capture.

    We show how the International Image Interoperability Framework (IIIF) Presentation API, combined with the Linked Open Usable Data (LOUD) design principles, provides a knowledge-representation layer capable of holding this ambiguity as a structural feature rather than resolving it. A custom IIIF Manifest uses the Choice construct to bind alternative material states to a single Canvas, enabling semantic annotation and metadata harmonisation across page-level variants without requiring separate records or lossy normalisation. We discuss how this approach extends to the W3C Web Annotation Data Model and Linked Art for distributed, provenance-aware enrichment (human and machine-generated), and how the resulting "boutique" Manifest raises concrete questions of long-term knowledge preservation: dependency on stable image-service infrastructure, the maintenance of dereferenceable URIs across system migrations, and the institutional labour required to keep interoperable resources trustworthy over time.

    By foregrounding IIIF/LOUD as an interoperable knowledge-management platform rather than a display technology, we argue that treating digitisation as interpretive modelling — rather than mass reproduction — offers a transferable pattern for other institutions managing materially complex, under-documented collections. We conclude with reflections on the trade-offs between the cost of purpose-built semantic modelling and its long-term reusability, provenance, and cross-collection interoperability.

  30. Expansion of Ad Hoc Abbreviations in Latvian Folk Song Variations Using Large Language Models

    Linda Barbare, University of Latvia, Latvia · Paper ID 46

    Abstract

    Since the 19th century, collections of Latvian folk songs have been published utilizing condensed, ad hoc abbreviation formats to optimize printing resources. This study presents an algorithmic methodology for the automated reconstruction of these abbreviated poetic variations into their complete, full-text forms. Focusing on a corpus of 45 101 folk song texts from Volume 13 of Latviešu tautasdziesmas “Latvian Folk Songs”, we converted the unstructured source data into a structured JSON format. Following the comparative testing of multiple Large Language Models (LLMs), Gemini 3.1 Pro was selected for the generative reconstruction task due to its ability of processing morphologically complex, under-resourced Latvian language and its dialects without hallucination. The model was given a prompt that was engineered using Role-Prompting, Instruction-Based Prompting, and Few-Shot Structure Definition frameworks to decode ad hoc abbreviations, such as line reordering and single-letter lexical substitutions. To optimize the manual validation process, there was also incorporated LLM self-elicitation to generate token-level confidence scores for each reconstruction. The model navigated both standard linguistic norms and specific dialectal nuances on lexical substitutions and structural realignments. While instances of deeply contextual, morphologically complex dialect abbreviations introduce minor edge-case errors, the study concludes that LLMs are highly effective for ad hoc abbreviation expansion in historical poetic texts of under-resourced language.

  31. Monogram Segmentation in Byzantine Seals using Deep Learning

    Saranga Kingkor Mahanta, Sorbonne Université, France · Paper ID 45

    Abstract

    Monograms on Byzantine seals are characterized by complex overlapping characters and frequent missing parts, making the task of manually isolating them from raw seal photographs quite laborious. This makes automatic segmentation essential for systematic and reproducible large-scale identification. We present a supervised learning framework for automatic stroke-level segmentation of Byzantine monograms from seal images, intended as a research aid for historians and sigillographers. We use 605 manually segmented examples from four Byzantine seal collections and compare a convolutional U-Net against a transformer-based SegFormer under multiple evaluation settings designed to assess both overall performance and robustness to variations in collection and image quality. SegFormer achieves a Dice score of 0.79 and a \textit{cl}Dice score of 0.88 when trained and evaluated using data from all four collections. It consistently outperforms U-Net across the evaluation settings, with the largest Dice improvement observed when generalizing to an unseen collection. We additionally deploy the system as a public web application that accepts raw seal photographs and returns predicted stroke masks, lowering the barrier to large-scale processing of seal collections.

  32. The proposed system is an AI-powered adaptive VR platform for exploring AlUla.

    Hessa Almutairi, Prince Sultan University, Saudi Arabia · Paper ID 39

    Abstract

    Many cultural heritage sites around the world are fragile and cannot be accessed freely due to safety and preservation

    concerns. This creates a challenge in allowing people to experience and learn from these important historical places.

    This research proposes an AI-powered AR/VR system that allows users to explore cultural heritage sites virtually without

    causing damage to the original locations. The system uses artificial intelligence to detect important features, guide users

    through the environment, and provide historical information in an interactive way.

    This approach helps protect heritage sites while making them accessible to a wider audience. In addition, it creates an

    engaging and educational experience for users. The study shows that combining technology with cultural heritage can

    improve accessibility, support preservation, and increase awareness for future generations.

  33. Exploring the Applicability of AI-Powered Storytelling Tools in Cultural Heritage Education: An Exploratory Study in the Cyprus Context

    Kholoud Fadel, Individual · Paper ID 35

    Abstract

    The intersection of artificial intelligence and cultural heritage (CH) education presents an emerging frontier that remains underexplored, particularly in small island contexts such as Cyprus. While AI-powered storytelling tools have gained traction in broader educational technology discourse, their specific applicability to CH learning environments, where historical depth, multilingual complexity, and community identity are central, has received limited scholarly attention.

    This paper offers an exploratory conceptual examination of how AI-driven storytelling tools may be applied within CH education in Cyprus, a territory of exceptional archaeological and intangible heritage significance situated at the crossroads of European, Middle Eastern, and Eastern Mediterranean civilisations. Drawing on emerging frameworks in AI-supported content creation, digital heritage communication, and educational technology adoption, the paper interrogates three interrelated questions: What categories of AI storytelling tools hold the greatest pedagogical promise for CH education? What contextual factors in Cyprus, including linguistic duality, post-conflict heritage sensitivity, and tourism-driven heritage commodification, shape the applicability of such tools? And what conceptual framework can guide future empirical inquiry in this domain?

    The paper contributes an original applicability framework, provisionally termed the Contextual AI Storytelling Readiness Model (CASRM), which maps tool capabilities against local heritage education conditions across four dimensions: narrative adaptability, cultural sensitivity, learner engagement, and institutional readiness. The Cyprus context serves as an instructive case precisely because of its complexity, a small but heritage-rich environment where digital transformation in education is advancing yet CH-specific EdTech remains nascent.

    This work is positioned as a foundation for future empirical investigation, inviting researchers and practitioners to test, refine, and extend the proposed framework across comparable Mediterranean and island heritage contexts. The paper argues that AI storytelling tools, when contextualised thoughtfully, hold significant potential to transform how cultural heritage is transmitted, interpreted, and experienced by future generations.

  34. An Integrated Hazard–Ecology–Perception Framework for Cultural Landscape Planning

    Jingsen Lian, Delft University of Technology (TU Delft), The Netherlands · Paper ID 34

    Abstract

    Cultural landscapes are increasingly vulnerable to the compounded effects of geographical risks, ecological degradation, and imbalanced heritage value perceptions under intensifying climate change and global urbanization pressure. However, planning frameworks rarely integrate geographical hazards, ecological sensitivity, and both expert and public heritage value perceptions, often leading to insufficient planning responses. This study proposes a Hazard-Ecology-Perception Landscape Planning (HEPLP) framework that links these three dimensions for cultural landscape planning and evaluates it through a case study of Chengde Mountain Resort. A GIS-based approach is used to analyze geographical hazards and ecological sensitivity, while heritage value perceptions are derived from expert scoring and large language model analysis. The results identify one cluster where the high-risk geological hazards require complete conservation. Another eight clusters are also identified where trade-offs between ecological sensitivity and heritage value perception are carefully balanced, revealing that public perceptions conflict with ecological limits, or that clusters with low ecological sensitivity hold undervalued potential for development. Building on these insights, HEPLP delivers cluster-targeted conservation and development strategies that can be directly translated into practical zoning guidelines. Conceptually, unlike previous frameworks that focused on single or dual dimensions, HEPLP advances landscape planning as an integrative tool for sustainable cultural landscape conservation and development under environmental and social challenges.

  35. From Digital Collections to Immersive Experiences: User-Centred Access to Cultural Heritage through the eKultura Portal

    Jelena Rubić, Cultural Heritage Digitization Office, Ministry of Culture and Media of the Republic of Croatia, Croatia · Paper ID 76

    Abstract

    Digital cultural heritage platforms are increasingly evolving from repositories of digitised content into user-centred environments that support interpretation, interaction, learning and reuse. The Croatian eKultura system and its public portal exemplify this development by bringing together the needs of cultural heritage institutions as content providers and citizens as users, while progressively expanding the ways in which digital heritage can be accessed and experienced.

    For cultural heritage institutions, the eKultura portal provides a range of services supporting the systematic management and presentation of digital collections. Registered institutions can submit and manage digital content, provide mandatory digital deposit copies for libraries, create and publish 2D virtual exhibitions based on their digitised collections. These functionalities enable institutions not only to contribute digital resources to the national cultural heritage infrastructure, but also to contextualise and interpret their collections for broader audiences.

    For citizens and other users, eKultura provides a single point of access to diverse digital collections from Croatian museums, libraries, archives and other cultural heritage institutions. Content can be explored through thematic, chronological and material-based categories, as well as through references to institutions, authors and historical sites. Users can access and download digital content, request high-resolution copies or save favourites, add personal notes and share selected content through social media. Curated content, including featured collections, educational materials, virtual exhibitions and virtual tours, provides multiple pathways for engagement and supports users with different interests, backgrounds and levels of prior knowledge.

    An important dimension of this approach is the transformation of digital heritage from individual digital objects into contextualised narratives and interactive experiences. Virtual exhibitions enable institutions to apply storytelling and interpretative approaches, allowing users to encounter cultural heritage through thematic narratives rather than solely through catalogue records and metadata. This approach is further extended through the eKultura AR exhibition, which uses 3D cultural heritage objects from the eKultura portal to create an augmented reality experience. Presented in cultural institutions at different locations across Croatia, the exhibition demonstrates how digitised heritage can move beyond the online environment and become part of physical exhibition contexts, connecting digital collections with forms of public engagement.

    Expanding its digital heritage services, eKultura has developed a 3D virtual gallery environment that enables institutions to create more spatial, interactive and immersive virtual exhibitions. This environment opens up new possibilities for the presentation of collections, contextual storytelling and the exploration of 3D cultural heritage objects. The eKultura experience demonstrates how a national digital heritage infrastructure can combine structured access to collections with user-centred services, interpretation and emerging immersive technologies.

  36. From Scientific Collections to Digital Memory: Challenges and Opportunities for Research Libraries in Azerbaijan

    Gunel Huseynova, ANAS Central Scientific Library, Azerbaijan · Paper ID 70

    Abstract

    Scientific libraries play a fundamental role in preserving, organizing, and providing access to documentary heritage. In the digital era, however, the preservation of scientific collections increasingly extends beyond the physical protection of books, manuscripts, periodicals, and archival materials. Digitization, digital preservation, metadata management, electronic catalogues, and emerging information technologies are transforming the ways in which scientific heritage is documented, accessed, and transmitted to future generations. This paper examines the challenges and opportunities associated with the transition from traditional scientific collections to digital memory, with particular attention to the context of research libraries in Azerbaijan.

    The study focuses on the experience and information resources of the Central Scientific Library of the Azerbaijan National Academy of Sciences (CSL ANAS) as a case study. As one of the major scientific information institutions in Azerbaijan, the library preserves collections reflecting different periods and fields of the country’s scientific and cultural development. These collections include books, periodicals, bibliographic resources, rare publications, and other documentary materials that constitute an important part of the national scientific heritage. Their systematic description, preservation, digitization, and integration into contemporary information systems are essential for ensuring long-term accessibility and supporting scholarly research.

    The paper discusses several challenges faced by research libraries in the process of digital transformation, including the preservation of fragile and rare materials, the selection of resources for digitization, metadata quality, interoperability, copyright considerations, technological sustainability, and the need for appropriate human resources and professional expertise. At the same time, digital technologies create significant opportunities for improving access to scientific collections and strengthening cooperation among libraries, archives, museums, and research institutions.

    Particular attention is given to the concept of digital memory as a continuation of traditional library preservation functions. Digital memory is considered not simply as a collection of digitized documents, but as an organized and sustainable information environment in which scientific resources can be discovered, contextualized, connected, and reused for research and education. The paper also briefly considers the potential contribution of artificial intelligence and related technologies to future library services, including metadata generation, information retrieval, semantic linking, and the discovery of historical scientific resources.

    The study argues that the digital transformation of research libraries should be understood as a long-term process combining physical preservation, digitization, information organization, professional expertise, and sustainable access. The Azerbaijani experience demonstrates the importance of developing integrated approaches that connect existing scientific collections with emerging digital heritage infrastructures. Such approaches can contribute to the preservation of national scientific memory while also increasing the international visibility and accessibility of Azerbaijan’s documentary heritage.

  37. Re-rooting Akhenaten's Legacy: Bridging the Provenance and Legislation Gap of Amarna Departed Treasures with a Data-Sensorium Framework into its Egyptian Context

    Mohamed Soliman, Center of Archaeological Studies (MoTA) Egypt & AAII, Musahsino University, Japan · Paper ID 58

    Abstract

    The Akhenaten Museum in the al-Minya Governorate displays the unique Amarna era, an outstanding period in ancient Egypt during the 18th Dynasty (1353–1336 BC). Following the religious conflict that escalated between Akhenaten and the influential priests of Amun in Thebes, Akhenaten established a new capital compatible with his doctrine: “Akhet-Aten” (The Horizon of Aten), known today as Tel al-Amarna. The Egyptian pantheon was restored under Akhenaten's successor, Tutankhamun, and the priests of Amun subsequently destroyed Akhet-Aten.

    The surviving treasures of Amarna are scattered across global museums. This dispersion was driven by early 20th-century legislative lacunae in Egypt and modern UNESCO regulatory constraints that prevent the repatriation of artifacts displaced prior to 1970. Despite the Akhenaten Museum of accessible location, it locates approximately 90 km north of the Tel al-Amarna site, the artifacts remain isolated from their authentic spatial and urban context.

    To overcome these geographic and legislative barriers, this contribution outlines a collaborative initiative between MoTA and AAII to implement the Data-Sensorium system within the museum. Data-Sensorium is an innovative framework designed to bridge the gap between digital data and human sensation.

    This contribution argues that visually integrating globally scattered Amarna treasures into an authentic cyber-urban environment can virtually re-root Akhenaten’s legacy within the physical museum. By adopting an interdisciplinary approach—which blends archaeology, digital storytelling, 3D modelling, and AI-driven tools—the initiative restores a sense of provenance and advances the "no-frontier museums" concept, allowing digital curation to transcend international and physical borders.

  38. 14:30–14:50

    Coffee Break

  39. 14:50–17:00

    Session 9: Extended Memories: XR, VR and MR in Digital Cultural Heritage

  40. User Profiling and Kano–QFD-Based Requirements Prioritisation for Digital Natural Heritage Services

    Yujin Yoon, Korea National University of Heritage, South Korea · Paper ID 211

    Abstract

    Natural heritage encompasses ecological and cultural relationships between humans and nature, whose contextual values should be considered in digital transformation [1]. However, fragmented biodiversity resources can hinder access to interconnected information [2]. The memory twin concept offers an approach to enriching digital representations with heritage knowledge, context, and meaning [3]. Translating these connections into meaningful exploration requires experience design that addresses users’ knowledge and purposes. This study develops user profiles for digital natural heritage services and identifies interface requirements and design priorities through Kano–Quality Function Deployment (QFD), an approach used in interface design to classify user requirements and translate them into design priorities.

    We reviewed literature on user characteristics, information needs, and exploration behaviour in museums and biodiversity data services [4,5]. This review informed an interview guide covering required information, exploration functions, and conditions for using digital resources. Drawing on expert interview methodology [6], we conducted semi-structured interviews with four experts working across natural heritage research, policy, exhibitions, and education. Interview transcripts were segmented into meaning units, and similar statements were consolidated into 13 core requirements. The requirements analysis and literature review informed the structuring of user context into four dimensions: Knowledge Profile (expertise and domain familiarity), Purpose Profile (intended use), Preference Profile (areas of interest), and Exploration Behavior. Requirements were then classified into Must-be, One-dimensional, and Attractive quality layers using the Kano model [7]. This classification interpreted the expected benefits of fulfilling each requirement and the constraints associated with its absence, as expressed in the interviews.

    Using QFD [8], we mapped the 13 requirements to 11 design attributes and calculated normalised priorities. Ranking relations and exploration paths emerged as the highest priority, followed by 3D analytical representation. Temporal modelling of heritage relations and modular assets shared the next priority level. We then aligned key functions with four exploration profiles: Research and Analysis, Content Creation, Citizen Participation, and Heritage Interpretation. Research and Analysis emphasised precise observation and evidence verification; Content Creation, modular composition and reuse; Citizen Participation, place-based access and verification of observations; and Heritage Interpretation, understanding ecological and cultural relations. This alignment distinguished common functions for checking data reliability and reuse conditions from exploration functions adapted to users’ expertise and purposes.

    The study provides design criteria for conveying the contextual values of natural heritage through interfaces that address differences in users’ information needs and exploration behaviour. Because the requirements were derived primarily from four expert interviews, they may not fully represent the needs of wider user groups. Future research should validate and refine the requirements through empirical studies with diverse users. The findings provide a basis for subsequent digital archive development, particularly the modelling of relations among heritage resources and the provision of provenance information to support assessments of reliability.

  41. Stronger Together: Connecting Europe’s Cultural Heritage through 3D, XR and AI Innovation

    Susanna Capannini, PHOTOCONSORTIUM, Italy · Paper ID 214

    Abstract

    By bringing cultural heritage into dialogue with advanced scientific research and technological innovation, EUreka3D-XR – European Union’s REKonstructed content in 3D to produce XR experiences (2025–2026) has advanced the digital transformation of the cultural heritage sector, building upon the foundations, infrastructure and achievements of its predecessor, the Digital Europe-funded EUreka3D project. Together, these initiatives establish a coherent pathway from the high-quality digitisation and documentation of cultural heritage assets to their meaningful reuse in immersive, interactive and educational experiences as it has been demonstrated in the two Pan-European campaigns of TwinIT Part I and TwinIT Part II. A defining strength of EUreka3D-XR lies in its ability to build upon the distinctive results of EUreka3D, including the EUreka3D Data Hub, its dedicated 3D Viewer, the supporting technical infrastructure, and a knowledge base of documentation, case studies, recordings, guidelines and recommendations for 3D digitisation. Particularly significant is the adoption of a quality-driven, paradata-based documentation methodology grounded in the European Commission’s VIGIE2020/654 study, which established a systematic approach to assessing and documenting the quality, reliability and transparency of 3D cultural heritage assets in combination with the FAIR and CARE guidelines. These foundations provide the methodological and technological continuity necessary to transform existing digital resources into trustworthy, reusable and impactful cultural heritage experiences.

    Building upon this established ecosystem, EUreka3D-XR delivered three innovative use cases demonstrating the potential of Extended Reality (XR): Virtual Girona: Discovering the Historical Walls, The AR Narrative of the Hidden Side of Bibracte, and A New Life for Saint Neophytos’ Enkleistra. The project also developed five digital tools to facilitate the transformation of existing 3D resources into engaging visitor experiences, alongside a substantial portfolio of educational resources, including Open Educational Resources (OERs), three online courses, and additional interactive learning activities. Collectively, these results demonstrate how validated digital assets, innovative technological tools and interdisciplinary expertise can enhance cultural heritage interpretation, accessibility, education and public engagement.

    A particularly important contribution is the introduction and practical exploration of the Memory Twin concept, marking a significant conceptual advance in the digital documentation of cultural heritage. Whereas a conventional Digital Twin primarily represents an asset’s geometry and associated physical, spatial and material characteristics, a Memory Twin extends this representation by incorporating verified narratives, historical context, collective memory, intangible heritage, lived experiences, identity and cultural values. This holistic perspective seeks to preserve not only the physical evidence of the past but also the meanings and human connections that make cultural heritage significant to present and future generations.

    Beyond technological development, EUreka3D-XR strengthened European capacity building through seminars, demonstrations, workshops and participation in events across Europe, continuing the training and knowledge-transfer activities initiated by EUreka3D. By promoting access to digital resources, supporting collaboration with external organisations, and contributing to initiatives such as the Twin it! campaign, the project encouraged the responsible reuse of cultural heritage assets within the Common European Data Space for Cultural Heritage. It also fostered the exchange of expertise among researchers, cultural heritage professionals, technology developers and institutions, supporting the wider adoption of 3D digitisation and XR workflows.

  42. From Digital Archives to Living Narratives: Repurposing High Resolution Textile Heritage Data for Immersive Cultural Learning Beyond Museums.

    Delas Santano, Sunway University, Malaysia · Paper ID 158

    Abstract

    The digitisation of cultural heritage collections has traditionally prioritised documentation, preservation and database archival accessibility. However, the increasing availability of high-resolution digital heritage assets presents opportunities to extend their value beyond institutional repositories into interactive, educational and publicly accessible environments. This paper investigates how high-fidelity digitisation can serve as the foundation for reusable interpretive experiences that communicate both the tangible characteristics and intangible cultural knowledge embedded within traditional Malay textiles.

    Drawing upon an ongoing research creation initiative undertaken for Malaysia's Merdeka Textile Museum, the study examines the transformation of digitised heritage assets into multiple forms of immersive cultural interpretation. The project encompasses the high-resolution documentation of more than 300 artefacts, including songket, tenun, batik and other traditional textiles and adornments. Imaging techniques including pixel shift multi-shooting, enable the capture of intricate material details, weaving structures and decorative motifs, establishing a digital resource for preservation, research and subsequent interpretive development.

    Adopting research through design approach, the paper examines how these digital resources inform the development of augmented reality applications, interactive motif exploration, tactile interfaces and projection-based storytelling. It considers their deployment and adaptation across museum exhibitions and everyday public environments, including shopping malls, where immersive technologies can introduce cultural knowledge to audiences beyond conventional heritage institutions. Attention is given to the relationship between visual fidelity, cultural interpretation, interaction design and the translation of complex heritage knowledge into accessible learning experiences.

    The study proposes a reusable digital heritage interpretation framework connecting five interrelated processes: high fidelity acquisition, digital asset enrichment, interpretive transformation, contextual deployment and reflective evaluation. Through this framework, digitised artefacts are understood not merely as archival representations but as evolving cultural resources capable of supporting multiple educational applications.

    The paper contributes to discussions surrounding the circular economy of knowledge in digital cultural heritage by demonstrating how carefully documented heritage assets can be repurposed across technologies and learning environments not just in the classroom and museums, but also across commercial and public spaces. It further considers the importance of maintaining cultural authenticity, interpretive integrity and accessibility as heritage content transitions between physical collections and digitally mediated public experiences. Ultimately, the research argues for a shift from digitisation as an endpoint towards digitisation as an enabling foundation for sustainable, meaningful and widely accessible cultural learning.

  43. A VR reconstruction of Tumuli, Rock Art and Sahrawi Sky Lore in Western Sahara

    Maitane Urrutia-Aparicio, West University of Timisoara and Universidad Europea de Canarias, Romania/Spain · Paper ID 170

    Abstract

    Western Sahara holds a dense and still poorly studied record of prehistoric dry-stone funerary monuments, together with engraved rocks with cupmarks, spirals and Libyco-Berber inscriptions. Fieldwork in the region is limited by remoteness, harsh conditions and the political dispute over the territory. Due to this, and in order to support research and as means of public engagement, we present a virtual reality (VR) scenario built in Unity 6 of one of the sectors surveyed in 2025. Users can walk around, interact with objects and discover more information on the natural and anthropic elements making up the region.

    For the ground layer we have used a digital terrain model (NASA & NGA, 2013) at 30 m resolution, with the tumuli placed at their real coordinates. This makes another way of showing the possible intervisibility of heritage elements and spatial analysis. We added photogrammetric 3D models that users can inspect up close, both for tumuli and engravings that we found on the field. Some 3D models for smaller objects such as pottery and lithics were generated from images taken on site using genAI models such as Rodin v2 (Zhang, Q., & Zhang, 2024). We incorporated sounds and some local flora and fauna to make it more realistic, as well as display the information available on them.

    For the sky we have reproduced a model for the latitude of the focus area with real time positions for the sun, moon and stars. Statistical analysis of hundreds of monuments in Western Sahara has shown that their orientations are consistent with the visibility of the crescent moon near the winter solstice (Rodríguez-Antón et al. 2023a). Since the sky can be set for any date, including the periods proposed for the building of the tumuli, a user can stand beside a tumulus and watch celestial bodies rise or set over the horizon. In addition, we incorporated the oral memory related to the Sahrawi constellations, star names and their sayings and stories, which were collected from elders in the Tindouf refugee camps by Rodríguez-Antón et al. (2023b). For nomads, the stars served to predict the weather, travel through the desert and practice their religion.

    With this reconstruction, users can interact and connect the material records with a living oral tradition. They can explore a vast area, walking through a landscape they would rarely have the opportunity to visit, and this includes refugee communities that have been away from their land for years. Similar VR experiences for rock art have shown strong engagement from both archaeologists and indigenous stakeholders (Cassidy et al. 2019).

    References

    Cassidy, B., Sim, G., Robinson, D.W., Gandy, D.: A virtual reality platform for analyzing remote archaeological sites. Interacting with Computers 31(2), 167–176 (2019). https://doi.org/10.1093/iwc/iwz011

    National Aeronautics and Space Administration, & National Geospatial-Intelligence Agency. (2013). Shuttle Radar Topography Mission (SRTM GL1) Global 30m [Dataset]. OpenTopography. https://portal.opentopography.org/raster?opentopoID=OTSRTM.082015.4326.1

    Rodríguez-Antón, A., Urrutia-Aparicio, M., Perera Betancor, M.A.: Archaeoastronomy and conflict: on the orientation of prehistoric funerary monuments in Western Sahara. Sustainability 15(3), 2005 (2023a)

    Rodríguez-Antón, A.: Astronomy from the exile. A study of the Sahrawi astronomical knowledge with project ‘Amanar: under the same sky’. In: Cultural Astronomy and Ancient Skywatching, pp. 35–43. Totem Studio (2023b)

    Zhang, Q., & Zhang, L. (2024). Rodin: 3D asset creation with a text/image/3D-conditioned large generative model for creative frontier. ACM SIGGRAPH 2024 Real-Time Live! [Article 8]. Association for Computing Machinery. https://doi.org/10.1145/3641520.3665304

  44. ROMULUS: Object-Centred and AI-Assisted Historical Storytelling for Museum Interpretation

    Jörn Stock, ROMULUS – Narrative Interpretation for Museums & Heritage, Germany · Paper ID 167

    Abstract

    Museums seek ways to communicate the significance of objects, sites and historical sources to audiences with different levels of prior knowledge. Short audiovisual narratives can make such material approachable, but the increasing use of generative artificial intelligence introduces questions about historical accuracy, trust and the boundaries between documentation and interpretation. This practice-based contribution presents ROMULUS, an independent production initiative that explores these questions through short historical storytelling demonstrations for museums and heritage settings.

    ROMULUS begins, wherever possible, with a tangible object, place or primary source rather than with technology itself. A typical story lasts approximately 60–180 seconds and is developed through a workflow comprising source research, selection of a narrative focus, scriptwriting, narration, image and sequence design, editing and sound production. Historical photographs and other archival material can be combined with AI-assisted visual reconstructions when original imagery cannot show the event or human context being discussed. The intended output is a self-contained audiovisual sequence suitable for a museum display, projection or visitor-operated media station. The project is offered as a creative production service, not as an automated AI platform.

    Several exploratory demonstrations illustrate different interpretative approaches. A medieval drinking horn becomes a starting point for a story about trust and social bonds. A Viking ship offers a perspective on travel, departure and life aboard. A historical telegraph instrument connects technical change with the people who operated it, while a Tesla-coil demonstration explores the human context of electrical innovation. A more recent experiment uses a reconstructed first-person portrayal of Marie Skłodowska-Curie to introduce the discovery of radium and the Nobel Prizes. The Curie story was produced in English and Polish and uses selected archival imagery alongside synthetic visual material.

    The latter demonstration makes a central ethical tension especially visible. A convincing historical voice or face may help audiences engage with a narrative, yet it can also appear to provide direct testimony that never existed. Similar concerns apply to synthetic documents, reconstructed environments and the treatment of uncertain historical detail. Responsible deployment therefore requires explicit identification of reconstructions, separation of authentic sources from generated imagery, appropriate permissions for archival assets and review by the commissioning heritage institution. These are design requirements and open questions, not problems assumed to have been solved by visual quality alone.

    This contribution shares ROMULUS as a developing production practice and invites discussion about where AI-assisted interpretation can add value, where it should be limited and how institutional oversight might be incorporated. No visitor-impact study or educational effectiveness claim is presented. Future pilot collaborations could assess historical trust, accessibility, engagement and practical implementation in real exhibition environments.

  45. OpenAppiaHBIMM-XR Multi Modal Informative Models and Levels of Knowledge for Communicating Heritage Infrastructure. The Appian Way, and the Mausoleum of Cecilia Metella node-section

    Marzia Gabriele, Politecnico di Milano, Italy · Paper ID 90

    Abstract

    This essay examines the Appian Way “ Regina Viarum” (Roman military commercial road of c.540 km) as a cultural and landscape heritage infrastructure recently listed on the UNESCO World Heritage List (2024), also thanks to a massive high quality digitization (Brumana, 2023) promoted by PAAA (the Archeological Park of the Appian Way) along 10 km, focusing on how its fragmented, multilayered remains can be explored through Open-Air WebXR, accessible by smartphone-on-site remote fruition to the public of visitors and inhabitants walking across it, enhancing its comprehension. This study presents an integrated methodology in which field surveys, historical and archaeological research, analysis of construction methods, iconographyies, HBIM modeling, and Extended Reality (XR) narratives are considered as parts of the same holistic process, aimed to documenting cultural heritage for preservation purposes but also to its transmission.

    Different Levels of Knowledge (LOK), rooted in different informative models and rendered through Extended Reality (XR), are proposed as vehicles for the informed communication of different levels of material remains, information, survey data and research. The Mausoleum of Cecilia Metella node and its adjacent section are used to test a method of research and dissemination that can be scaled up to the broader system of the Appian Way. The Appian node-section models build on the 19th-century concept of an open-air museum, conceived by Luigi Canina, bridging the past and the present for a more effective relationship between heritage, information and the public. A Web-XR experience that unlocks knowledge of the Appian Way is presented, based on the integration of bibliographic, archival and iconographic sources with digital survey data. The information collected is structured within an HBIMM (multimodal Heritage BIM) 3D model that grows progressively as knowledge is acquired. It is intended not merely as a simple three-dimensional representation, but as a useful tool that links geometry to materials, building techniques, documentary sources and multiple historical phases, supporting PAAA preservation actions and communication.

    A dual mode of immersive journey was designed, reconstructing both the Cecilia Metella Mausoleum hyper-node and the Appian Way section near the node: a high-resolution, high-LOD/LOG HBIMM node, and a low-LOG virtual 3D reconstruction of the section in Roman times, starting from Canina’s ideal representations and verified against the almost invisible surviving traces.

    The node model was built by integrating terrestrial laser scanning, mobile mapping systems and photogrammetric data processed through 3D vectorial, NURBS, and HBIM procedures, ultimately resulting in a parametric model capable of incorporating all the monument’s informative dimensions. To make the knowledge accessible to a broader audience, the Metella LOD400 paradata describing the quality model (Brumana, 2024) were enriched with the HBIMM model node and transferred to the Twinmotion virtual environment. The visual output was enhanced with materials and textures from the photogrammetric surveys, narrative animations and XR informative navigation. The richness of the informative quality model supported the development of three narratives: the external marble layer and structure, the hypothesis of the concrete core arrangement, and the network of cunicula in relation to the adjacent lava flow.

    The Appian Way was, and still is, not perceived in its magnificence and uniqueness. As for the context, the Metella segment was surveyed and imagined by Luigi Canina in two representations, the “Survey” and the “Idea of the Appian Way” (Canina, 1853). The “Idea” perspective became the starting point for a virtual 3D geospatial XR reconstruction. Following a comparison between this ideal reconstruction and the current state, as documented by the digitization of the Appian Way (Brumana, 2023), the digital environment was exported to support informed, immersive geospatial experiences (Gabriele, 2024). The surrounding XR context elements (such as Roman sepulcreta) were generated with the support of generative AI as low-LOG models, then validated against the survey data and the surviving remains. Despite the low LOG, cross-referencing with the stratigraphy of the current Appian Way at foundation level, now covered (Paris, 2020), and with the circular portion of Valadier’s reconstruction of the main façade, allowed a dimensional check of the AI model. This revealed a mismatch between the hypothesized position of the Tomb of Titus Crustidius Briso and its actual position.

    The transition from scientific literature to virtual communication has, however, revealed several technical and methodological challenges, such as the processing of exceptionally complex geometries, the application of “decals”, model performance and, above all, the limitations of AI-based 3D generation tools and of cloud-based Web-XR resizing (Banfi, 2021).

    The research highlighted how crucial it is to maintain critical oversight throughout all stages of the process. Visitors must be able to distinguish between information based on verified data and the more speculative sources introduced for narrative purposes in the ideal reconstruction. To this end, the virtualization is presented as a time-lapse: it starts from the current remains and progressively grows back into the past.

  46. Immersive Virtual Reality and Serious Games for Digital Cultural Heritage: The TOURAL Approach to Underwater Exploration and Interactive Heritage Learning

    Efstathia Martinopoulou, Centre for Research & Technology Hellas (CERTH) /Information Technologies Institute, 57001, Thessaloniki, Greece · Paper ID 199

    Abstract

    Digital cultural heritage technologies offer opportunities to overcome physical and geographical barriers to heritage access while supporting interactive learning and cultural engagement. However, archaeological sites, particularly underwater heritage sites, remain challenging to access due to environmental conditions, specialised equipment requirements and the diving expertise needed for physical exploration. By bringing together Virtual Reality (VR), serious games and cultural heritage interpretation, this paper presents the design and implementation of two complementary applications developed within the Horizon Europe TOURAL project: a VR Diving Application and an object-based Serious Game.

    The VR Diving Application enables immersive dry-diving exploration of underwater cultural heritage sites. It leverages three-dimensional reconstructions of underwater landscapes and archaeological artefacts, combined with realistic environmental simulations, including underwater lighting, caustic effects, light refraction, spatial audio and dynamic representations of marine flora and fauna. Interactive Points of Interest and live underwater camera streaming support interpretation.

    The Serious Game complements virtual exploration through the interactive reconstruction of digitised cultural artefacts. Three-dimensional object models are prepared and their geometry fragmented into puzzle pieces. Users manipulate the scattered pieces to reconstruct the original artefacts. Contextual information and gallery views connect puzzle completion with the historical significance of the objects.

    Together, the applications illustrate a modular approach to turning digital heritage assets into interactive experiences that combine virtual access, contextual storytelling and game-based participation. Their reusable technical foundations can accommodate additional sites, objects and educational scenarios, offering opportunities for heritage interpretation through museums, visitor centres and other local cultural spaces, particularly in rural and remote destinations.

  47. Beyond Digital Preservation: A Three-Space Immersive Approach to Living Silk Heritage

    Hatefeh Majidi, Freelance · Paper ID 197

    Abstract

    This practice-based research explores how virtual reality can integrate spatial documentation,

    documentary evidence and artistic interpretation to communicate tangible and intangible dimensions of

    living silk heritage. Silk Heritage in Virtual Reality is an ongoing immersive project developed in Unreal

    Engine, organised around three interconnected environments: a documented silk workshop, a poetic

    performance space and a planned historical memory room.

    The case study emerges from fieldwork at Tecnoseta in Zagarolo, Italy, a contemporary silk production

    setting. Its machines, production procedures, materials and working environment raise a central question:

    how can digital heritage experiences convey situated knowledge and human experience rather than

    reproducing only the appearance of objects and places?

    The first environment, The Living Workshop, draws on photogrammetry and Gaussian Splatting captures

    made inside the actual production space. Its intended interaction design allows visitors to explore the

    digitised workshop, approach selected machines and activate documentary footage explaining their

    functions and associated production processes. The workshop capture has been undertaken; the

    interactive navigation and film-triggering features are part of the proposed experience rather than a

    completed system.

    The second environment, The Moth Room, uses the language of theatre within VR. Narrated by a

    domesticated silkworm moth unable to fly, this space explores transformations between living bodies,

    human labour and cultural material. Scanned silk-related objects, recorded video, directed lighting,

    darkness, spoken narration, music and environmental sound form a sequence of staged encounters. A

    substantial portion of this environment has been assembled in Unreal Engine, including the narration,

    audiovisual composition and integration of two scanned objects. Its experimental design tests how artistic

    interpretation may complement factual documentation without being mistaken for it.

    The third environment, The Memory of Silk, remains at the conceptual stage. It proposes to connect the

    contemporary workshop and the moth's first-person account with the broader cultural history of silk

    through archival photographs and moving images. AI-assisted animation of selected historical stills is

    being considered, with an emphasis on clearly differentiating archival sources from generated

    interpretation.

    Together, these spaces propose a multi-layered approach combining documentation, embodied narration

    and cultural memory. Rather than reporting evaluation outcomes from a finished VR experience, this

    work-in-progress identifies practical and interpretive questions emerging from production: how to

    connect volumetric capture with documentary film, how to represent working gestures that are difficult to

    capture spatially, and how to preserve transparency between recorded evidence and fictionalised

    storytelling. The project contributes an artistic research case study for exploring digital heritage as an

    experiential form of knowledge transmission.

  48. Short Discussion

  49. Beyond Digital Reconstruction: BIM, HBIM and HHBIM for Sustainable Heritage Management

  50. Information – from Heritage Description to Heritage Constitution

    Drazen Arbutina, Zagreb University of Applied Sciences (Tehničko veleučilište u Zagrebu – TVZ), Croatia · Paper ID 198

    Abstract

    Information has always been of crucial importance for the conservation of heritage. Ontologically, heritage has always been a bearer of information. In the case of intangible cultural heritage, this information was also an independent bearer of value; it was encoded as a set of various rules and customs that defined the cultural identity of individuals and communities. In the case of tangible heritage, it is precisely information that ensured the description of values and their attributes, ultimately, information also enabled the dissemination of knowledge regarding its existence. In this sense, information related to heritage has always been consistent with the era in which it was created—that is, the era in which it was used. The intention is to demonstrate that, over time, the most modern methods available for collecting, processing, and ultimately disseminating heritage information have always been utilized (as an immanent principle). Therefore, the purpose of this paper is, on the one hand, to address the phenomenology of information gathering/collection, information transformation and interpretation, as well as their systematization, storage, and ultimate dissemination alongside an analysis of various material, technical and technological limitations, possibilities and expectations, especially those related to architectural heritage. The intention is to highlight the specificities of historical shifts and technological developments within them, as well as the constant relationship toward available information which—regardless of the method of collection—must be adequate for the protection and preservation of heritage.

    In this context, the development of information collection methodology and technology will be addressed, ranging from classical analog methods, which were highly time-consuming, to modern methods of digital information collection (from individual and partial approaches to real-time ones) utilizing various techniques, including cameras and infrared sensors, together with LiDAR scanning. This will be accompanied by examples defining the relevance and appropriateness of modern techniques, technologies, and specific equipment for various forms of heritage, primarily architectural heritage not just in the area of gathering but also in the area of information transformation regarding the model Level of Development (LOD) consisting of level of geometry and level of information.

    Accordingly, the intention is also to present contemporary possibilities for collecting, processing, storing, and transmitting information about building/architectural heritage. Furthermore, the paper aims to discuss information issues within the context of contemporary conservation of tangible and intangible heritage. This will be explored in the context of conservation principles for preserving the integrity and authenticity of recognized values and their material and other attributes, as well as the principles of preparing and processing collected information, with reference to the context and content associated with the terms of BIM (Building Information Modeling together with Management of acquired information), HBIM (Heritage Building Information Modeling/Management), Digital Twin, and Memory Twin on the related examples of building/architectural heritage within Croatia (recent experiences with reconstruction and rehabilitation following war devastation and natural disasters, such as earthquakes or fires, as well as matters related to risk preparedness and risk mitigation in general).

  51. Stakeholder Acceptance of Digital and Information Technologies in Cultural Heritage Buildings: A Reflexive Thematic Analysis of the Greek INHERIT Pilot

    Eleni Manolopoulou, Singular Logic, Greece · Paper ID 172

    Abstract

    Digital and information technology solutions offer opportunities to improve cultural heritage conservation and building management. Their practical value, however, depends on how stakeholders understand, accept, and sustain them within protected environments. Existing research often foregrounds technical capabilities, leaving a need for situated accounts of the organizational, social, and heritage-specific conditions shaping adoption. This qualitative case study examines how stakeholders involved in the EU-funded INHERIT project’s ELLET pilot in Athens, Greece, perceive and accept these solutions, and what facilitates or hinders their adoption. Particular attention is given to how the building’s geographic and cultural setting enters stakeholders’ accounts. Six semi-structured interviews were conducted with two heritage professionals, three technology providers, and one public authority stakeholder. Interviews were analyzed using reflexive thematic analysis. Four themes were developed: modernization is acceptable when it sustains heritage and everyday use; data earns value by making building care targeted and anticipatory; adoption must fit people as well as buildings; and installing a solution is not the same as sustaining it. Across the accounts, acceptance was conditional on technologies supporting conservation and continued building use while respecting historic fabric, appearance, and heritage meaning. Monitoring was valued for making environmental conditions intelligible and informing targeted, preventive decisions. Its usefulness depended on reliable measurements and interpretation alongside conservation and management expertise. Understandable outputs, manageable learning demands, compatibility with existing workflows, trust, and collaboration supported acceptance. Costs, conservation restrictions, limited expertise, and uncertainty about continuing maintenance and responsibility constrained implementation and sustained use. Participants differed in their assessments of affordability and implementation difficulties, indicating that these conditions were experienced unevenly across professional positions. The geographic and cultural setting shaped accounts through protection requirements, building use, local heritage meanings, and established stakeholder networks. The interviews offered limited grounds for attributing differences in acceptance primarily to national culture. As a small professional sample, the study provides contextual insight rather than population estimates or direct evidence of wider community acceptance. Reported benefits also reflect stakeholder assessments rather than independently measured technological outcomes. The findings position acceptance as an ongoing, situated process extending beyond favourable attitudes or initial installation. For INHERIT and comparable heritage initiatives, they highlight the importance of heritage-sensitive design, accessible and actionable information, collaborative implementation, and early planning for technical support, funding, and organizational responsibility beyond project completion. The study contributes an account of how perceived technological benefits become acceptable and potentially sustainable within the practical conditions of heritage-building care.

  52. Beyond Digital Documentation: Towards a Conservation-Led Heritage Digital Twin for Asinou Church, Cyprus

    Catharine Arul Dass, Middlesex University, London, United Kingdom · Paper ID 157

    Abstract

    Advances in Digital Cultural Heritage have created new opportunities for documenting, analysing and interpreting historic buildings. However, extensive digital information does not necessarily form an integrated knowledge environment capable of supporting long-term conservation and lifecycle management. This paper examines the UNESCO listed Church of Panagia Phorbiotissa, commonly known as Asinou Church, Cyprus, as a critical case study in the transition from advance digital documentation towards a Conservation-Led Heritage Digital Twin (CL-HDT). The study forms part of wider PhD research.

    Asinou Church has been extensively recorded and studied through terrestrial laser scanning, photogrammetry, Heritage Building Information Modelling (HBIM), semantic enrichment, structural analysis and digital dissemination. Together, these outputs provide a substantial digital evidence base. However, a persistent gap remains between the production of technically sophisticated digital documentation and the effective integration and use of its data for conservation planning, risk management, condition monitoring and long-term stewardship.

    An interpretive, evidence-led gap analysis accesses the extent to which existing digital resources support heritage lifecycle requirements. It examines the representation and integration of asset identity, geographical context, physical fabric, heritage significance, condition, deterioration, risk, intervention history, documentary evidence, provenance, ground conditions, continuous use, function, memories and uncertainty. The analysis finds that these information categories are often dispersed across separate models, records and repositories, with limited semantic connections. As a result, links between evidence, professional assessment, conservation decisions, authorised actions and subsequent monitoring are not consistently recorded or updated.

    The findings, show that geometric accuracy and object level semantic enrichment, while fundamental, are insufficient to create an operational conservation-led knowledge environment. Existing advanced HBIM models primarily document the asset at a particular point in time, offering limited support for tracing changes in condition, cultural significance, function, risk and conservation requirements throughout its lifecycle. Further gaps relate to provenance, confidence, accountability and continuity. Sources, authorship, supporting evidence and levels of certainty are not always clearly linked to the relevant heritage component or conservation decision, reducing transparency and limiting future understanding and reuse. Knowledge gained through inspection, intervention and post intervention evaluations is also not consistently fed back into the evolving digital record.

    The paper proposes that a CL-HDT should move beyond technological integration and real-time replication. It should be guided by conservation purpose, heritage significance, and accountable decision-making, allowing the digital record to evolve alongside the physical asset while preserving the evidence behind change. Asinou Church demonstrates the potential for advanced digital documentation to support integrated lifecycle knowledge management. The identified semantic, resilience, and governance gaps, provide a basis for future research into connecting fragmented datasets, incorporating Memory Twin and enabling trustworthy, conservation-led lifecycle management.

  53. From HBIM to Memory Twin: An Interactive Web and VR Immersive Visit of the Acropolium of Carthage

    Nadia Bouzgarrou, ENSA Nantes - ENAU Tunis (France & Tunisia ) · Paper ID 140

    Abstract

    Heritage documentation is going digital: laser scanning, photogrammetry, Gaussian Splatting and three-dimensional models are progressively replacing graphic archives. Within this shift, Heritage Building Information Modelling (HBIM) has become the reference framework for structuring knowledge about a building. Holistic HBIM (HHBIM) widens it to historical and documentary dimensions, and the Memory Twin extends the digital twin towards the memory of places, that is, what has been lived there and not only what stands there. In parallel, virtual visits and XR devices make such models accessible to a wide public.

    Yet HBIM remains a document of geometry, materials and construction phases, not of the sensory memory of a place. The light that changes with the hour and the season, the smoke of incense, candle flames, voices, bells and reverberation are absent from it. Capture-based approaches (Gaussian Splatting, 360° tours) restitute a place as it was on the day of the survey, without allowing its conditions to be changed.

    This work belongs to an ongoing doctoral thesis in joint PhD, within the BIMaSPHERE project. Its case study is the Acropolium of Carthage in Tunis, Tunisia (former Saint-Louis Cathedral), modelled in IFC4 at LOD 300 from a 3D laser scanning survey: 696 objects, including 60 oriented windows and 19 spaces. From this model we developed an interactive prototype of immersive visit that runs in a web browser and in a VR headset (WebXR). The visitor walks in first person through the 19 spaces and composes the ambiance: (1) sunlight, computed for Carthage by hour and season, with the building axis set to its surveyed azimuth of 120.0°, cast through windows whose glass can be changed; (2) fog standing for incense, and candles; (3) spatialised sound (organ, chant, bells) with four trial reverberation times, from 1.2 to 8 s; (4) four predefined scenarios and an A/B switch to compare two ambiances at the same spot. The HBIM can be textured from 77 photographs projected onto its IFC elements, or superposed on the laser point cloud (median deviation from the model: 5.6 cm).

    The interest of the approach is that the ambiance is derived from the same model as geometry and semantics. Each parameter remains linked to an element (an opening, a surface, a space) and thus forms the ambiance layer of the HHBIM, the third layer of the Memory Twin. The limits are stated plainly: sounds are synthesised through survey in situ, reverberation times are trial values, and the model is not yet decorated.

    Evaluation is presented as a protocol, with no results at this stage. At the seven stations of the visit, participants will rate the place on six bipolar scales (−3 to +3), choose among six emotions and comment freely, while the application logs station, position, scenario, hour, season, settings and path. Comparing ratings across conditions will test which parameters change the perceived ambiance.

  54. Towards a Memory Twin: Preliminary Knowledge Organisation Requirements for Architectural Heritage in Japan

    Noriki Sakamoto, Nakasha Creative Corporation, Japan · Paper ID 119

    Abstract

    As post-war modern architecture in Japan ages, decisions concerning renovation, seismic upgrading, replacement, or demolition become central to building management. At the same time, records related to these buildings—drawings, photographs, reports, survey data, and architects’ archives—may be distributed across different organisations, collections, and media. The challenge is therefore not only to digitise such materials, but also to preserve the relationships between the physical building, changes over time, documentary evidence, interpretation, and the processes by which digital records are produced.

    This research was prompted by the digital archive project for the Former Hashima City Hall in Gifu Prefecture, designed by Junzo Sakakura and completed in 1959. Before demolition, the project recorded the building through spatial documentation and 360-degree VR, while drawings, historical photographs, and architect-related materials held by different institutions were brought together for online public access. The experience showed the value of collecting heterogeneous records before the physical building is lost, while also raising a further question: how can a digital archive move from a collection of resources to a structured knowledge environment in which the basis of interpretation remains traceable?

    To examine this question, the study reviews international research on Historic Building Information Modelling (HBIM), Digital Twins, Memory Twins, paradata, and provenance, with particular attention to structured information management, temporal change, and traceability. In parallel, it analyses the 2026 guidance of Japan’s Agency for Cultural Affairs on conservation and utilisation plans for cultural property buildings. Although the guidance was not developed as a digital information standard, it systematically addresses cultural significance, conservation and intervention histories, and the classification of buildings into zones (*bubun*) and components (*bui*) according to heritage value and permissible intervention. These practices are therefore examined as a source of domain-specific information requirements rather than as an HBIM specification.

    The comparison yields six preliminary requirements for knowledge organisation: (i) persistent identification of buildings, spaces, and components; (ii) event-based recording of change over time; (iii) explicit linkage between evidence and interpretation; (iv) provenance and paradata describing how records and models were produced; (v) cross-repository linking through persistent identifiers; and (vi) epistemic distinction between observed, documented, and interpreted information. These requirements are examined against representative record types from the Former Hashima City Hall archive to consider how spatial records, drawings, photographs, and historical sources can remain connected without requiring a fully implemented knowledge graph or Memory Twin.

    The study does not propose a complete Heritage BIM framework for Japan or claim to have implemented a Memory Twin. Instead, it identifies preliminary knowledge-organisation requirements for moving from digital archive collections toward structured and verifiable heritage information environments. This provides a practical basis for future development from digital archives and HBIM toward Memory Twin approaches for modern architectural heritage in Japan.

  55. From Documentation to Decision: An HBIM-Based Framework Integrating Value Engineering, SDGs and Multi-Criteria Analysis for Portuguese Colonial-Era Heritage in Timor-Leste

    Afranio Freitas, Beira Interior University, Portugal · Paper ID 223

    Abstract

    Built heritage from the Portuguese colonial period in Timor-Leste is an important cultural resource, yet its rehabilitation is constrained by deterioration in a humid tropical climate, limited documentation and scarce resources. Heritage Building Information Modelling (HBIM) is now an established approach for documenting historic buildings. Its potential as a knowledge-management and decision-support environment for rehabilitation planning, however, remains insufficiently explored, particularly in low-resource contexts where heritage, technical, economic and sustainability considerations must be reconciled through a transparent procedure.

    This research proposes an HBIM-based framework that integrates Value Engineering (VE), Sustainable Development Goal (SDG) assessment and Multi-Criteria Decision-Making (MCDM) to support rehabilitation decisions for this heritage. Following a design-science research approach, it develops a methodological artefact that connects heritage documentation, condition assessment, economic evaluation, sustainability assessment and stakeholder preferences in a single traceable decision chain.

    HBIM serves both as the knowledge-management environment and as the decision-support backbone of this chain. Data acquired through drone photogrammetry, total-station surveys, manual measurement, direct inspection and photographic documentation are structured so that each building element is linked to its condition records, quantities, intervention alternatives and supporting evidence. Observed data, expert interpretation and decision preferences are kept explicitly distinct, allowing every ranking to be traced back to its source records. Where essential data are missing, they are flagged rather than replaced by assumptions.

    On this information base, intervention alternatives are compared using four criteria: heritage significance, a VE factor, social sustainability and environmental sustainability. The two sustainability criteria are operationalised through research-specific indicators aligned with SDGs 7 (Affordable and Clean Energy), 11 (Sustainable Cities and Communities) and 13 (Climate Action). Criterion weights are derived from stakeholder preferences, so that institutional and conservation priorities are expressed explicitly.

    The main innovation is an admissibility-first decision rule. Before any multi-criteria aggregation, each alternative must be compatible with heritage significance and supported by complete essential technical data. Economic or sustainability advantages therefore cannot offset actions that compromise heritage values. Conservation becomes a precondition of the decision rather than one criterion traded against others.

    The framework is designed for a corpus of ten public heritage buildings across four municipalities of Timor-Leste, with its full application planned for the former Aipelo Prison as the pilot case. The pilot is intended to examine operational feasibility and contextual adaptability rather than to support statistical generalisation.

    The contribution to Digital Cultural Heritage is twofold. First, the framework extends HBIM from documentation towards an evidence-to-decision environment. Second, it formalises explicit conditions under which multi-criteria aggregation is admissible for heritage assets. For heritage authorities, it offers a decision procedure whose evidence, exclusions and rankings remain open to inspection and revision as new information enters the model. Although developed for Timor-Leste, the framework responds to conditions shared by many heritage contexts with limited documentation, restricted resources and climate-related deterioration. It is therefore proposed as a transferable reference for sustainable heritage rehabilitation and digital heritage management.

100% online · Free registration

Join EuroMed2026 online, wherever you are

Registration is free and open. Book your place on Crowdcast to participate in the online EuroMed2026 conference.

Book your free place(opens in a new tab)