Integrated data, metadata, and paradata management system for 3D Digital Cultural Heritage objects: Workflow automation, federated authentication, and publication

1. Introduction

• Large 3D files strain online delivery and client-side rendering, while paradata lacks established standards and 3D storage demands intensify network and processing constraints.The paper notes that multiple model versions can serve different use cases.Source: §1 · p. 1
• Europeana sharing requires support for diverse formats and access needs, online visualization, EDM-aligned metadata, and comprehensive acquisition and modeling paradata.Source: §1 · p. 1
• Ad hoc services and systems create duplicated work, redundant storage, and inefficient coordination across providers, tools, and technologies.Source: §1 · p. 2
• The proposed solution combines data and metadata management, automated processing workflows, federated authentication and authorization, and publication of resulting objects to Europeana.Its stated novelties include managing metadata and paradata with the data platform, automating their processing, and incorporating EDM.Source: §1 · p. 2

2. Problem Statement and Related Work

• A CHO comprises the 3D model and associated metadata and paradata, so its management must span ingestion through publication.The digitization process is multi-stage and can vary with infrastructure, software, and personnel.Source: §2 · pp. 2–3
• Existing digital infrastructures include repositories, aggregators, 3D viewers, and virtual research environments, with different roles in storing, compiling, visualizing, or supporting research.Source: §2 · p. 3
• The authors identify a lack of infrastructure for hosting 3D cultural heritage resources using tools and standards already supported by Europeana and the EU.Source: §2 · p. 3
• Onedata was selected to manage distributed CHO data and metadata with integrated workflows; prior uses include more than 4 PB of museum archival data and life-science data aggregation and processing.Source: §2 · p. 3

3. Cultural Heritage Objects Processing

• After digitisation, the proposed workflow uses authenticated upload, automated processing, and publication; documenting downstream processing supports reproducibility and reuse across objects.The digitisation itself remains the content provider’s responsibility, and workflows may be tailored to an object type or instance.Source: §3 · pp. 3–4

3.1. Federated Authentication and Authorisation

• EGI Check-in provides federated authentication and authorisation, connecting CHI identity-management solutions to secure content-provider uploads.It connects users, authentication servers, and services, and supports single sign-on.Source: §3.1 · p. 4

3.2. Data and Metadata Processing

• The processing workflow initializes a repository structure, validates uploaded models and metadata, ingests metadata, and creates smaller derivative models for publication.Validated objects and derivatives can be published as public unlisted shares, annotated with CHO metadata and processing paradata, and viewed in a browser.Source: §3.2 · p. 4

3.3.1. Europeana Data Model

• Europeana publication requires interoperable metadata and online 3D access, while EDM’s evolving support for structured paradata increases curation demands.Paradata documents digitisation, methods, and post-processing; the passage says it currently lacks a structured description mechanism.Source: §3.3 · p. 4, §3.3.1 · p. 5

3.3. Cultural Heritage Object Publication

• The publication design combines data ingestion, metadata and paradata management, automated processing, PID/DOI minting, and public advertisement through protocols such as OAI-PMH.These functions must comply with external standards for interoperability with Europeana.Source: §3.3 · p. 4

3.3.2. Onedata platform for integrated CHO management

• Onedata enhancements include a dynamic EDM editor, system-wide support for the model, and automated ingestion-to-publication through a REST API.A custom script executed the entire process programmatically for the Bibracte collection of about 500 3D models.Source: §3.3.2 · pp. 5–6

3.4. EGI DataHub

• The EGI DataHub deployment demonstrated an integrated path from ingestion and metadata curation to PID assignment, OAI-PMH advertisement, and Europeana harvesting.The proof of concept accommodated CHOs that became publicly available and discoverable on Europeana with 3D visualisations.Source: §3.4 · p. 6

4. Automation workflows with Onedata

• Integrated workflows run within Onedata, where users compose task sequences on selected collections and retain control of the lambdas’ processing logic.Lambdas can access Onedata APIs, and the execution environment is containerised and can use other task-execution frameworks beyond the current OpenFaaS implementation.Source: §4 · p. 6
• Onedata runs each lambda in a Kubernetes pod alongside a sidecar that mounts the Onedata filesystem for local filesystem access.Multiple pod instances can be deployed on demand to match workload.Source: §4 · pp. 6–7
• Workflow schemas sequence lambdas over typed stores, with lanes processed sequentially and parallel boxes allowing concurrent execution.Stores can hold initial user inputs, intermediate products, or final results.Source: §4 · p. 7
• Inventories let user groups share and reuse workflow definitions, with privileges controlling who can create, modify, or run them.The schemas are intended to hide internal task complexity from end-users.Source: §4 · p. 7
• The Figure 5 example calculates file checksums, stores file-checksum pairs, then writes a summary and assigns checksum metadata to each input file.The checksum store acts both as an intermediate container and as an execution artefact.Source: §4 · p. 7, Figure 5 · p. 7

• EUreka3D responds to a sector-wide need for modernised processes amid ambitious EU 3D digitisation targets and limited shared standards, knowledge, methodologies, infrastructure, and processes.Source: §5 · p. 7
• EUreka3D brought together content providers at different stages of 3D transformation, from controlled environments to complex objects in uncontrolled environments.Source: §5 · p. 7
• The project digitised a diverse range of objects according to technical specifications and accompanied models with metadata and paradata documenting the represented objects.Source: §5 · pp. 7–8
• The workflow has three phases: capture creates data, metadata, and paradata; cloud infrastructure stores them with tiered access; delivery prepares models and metadata for Europeana publication.Source: §5 · p. 8
• The implementation combines workflow automation, integrated data, metadata, and paradata management, federated authentication and authorization, Europeana publication, and distributed Data Hub infrastructure.Source: §5 · p. 8

• Evaluation covers a small movable figurine, the medium-sized immovable Lambousa Fishing Trawler, and the large complex immovable Church of the Holy Cross.Source: §6 · p. 8

• High-fidelity digitisation combines geometric and photometric data with metadata and paradata, supporting preservation, varied applications, and wider access across stakeholder communities.Source: §6.1 · p. 8
• Table 1 presents 3D data statistics for the Lambousa Fishing Trawler.Source: Table 1 · p. 9
• Manual preparation uses cultural-heritage experts for each digitisation step and prioritises quality, while computer-science experts automate post-processing to generate models at varied quality and size.Source: §6.1 · p. 9
• Each object's Europeana record links to a working 3D model in the EUreka3D viewer and includes detailed metadata and paradata.Source: §6.1 · p. 9

6.2.1. Historical Background

• Lambousa is a 1955 wooden liberty-style vessel and the last remaining trawler of its kind in the Panhellenic region; it was dry-docked for conservation in 2020.Source: §6.2.1 · p. 9

6.2.2. Digitisation Rationale

• The restoration documentation provides the 3D model in formats including 3DM, OBJ, STL, E57, and GLB, with paradata following VIGIE 2020/654 and metadata aligned with EDM.Source: §6.2.2 · p. 9

6.2.3. Digitisation Process

• The Memory Twin approach combines data, metadata, paradata, and intangible maritime knowledge; its photogrammetric survey processed 1,100 images into a georeferenced mesh with 27,363,867 faces.Source: §6.2.3 · p. 9
• The team aligned TLS and photogrammetric data to build a closed NURBS CAD model at Level of Detail 400, with deviation from the point cloud between -10 and 12 mm.Source: §6.2.3 · p. 10
• Intangible heritage was documented through archival research and interviews, including narratives from archaeologists, craftsmen, and engineers involved in the project.Source: §6.2.3 · p. 10

6.3.1. Historical Background

• The collection comprises approximately 1,500 archaeological discoveries spanning more than 7,000 years, from the Stone Age through Imperial Rome.Most displayed artefacts derive from the 1927–1931 Swedish Cyprus Expedition, which excavated at more than twenty sites.Source: §6.3.1 · p. 10

6.3.2. Digitisation Rationale

• The Swedish Cyprus Expedition Digitisation Initiative began in January 2024 under a bilateral agreement between the Cyprus University of Technology and Medelhavsmuseet.The agreement aims to digitise the largest collection of Cypriot antiquities outside Cyprus.Source: §6.3.2 · p. 10

6.3.3. Digitisation Process

• The digitisation followed the VIGIE 2020/654 Study and included paradata, metadata, and 2D/3D data.Small artefacts were photographed for photogrammetry, while larger immovable artefacts were scanned by hand; 31 artefacts were digitised.Source: §6.3.3 · pp. 10–11
• Twenty artefacts were aggregated through the National Aggregator of Cyprus, and two were aggregated through EUreka3D.The latter included a 3D model of one of the sculptures.Source: §6.3.3 · p. 11, Figure 8 · p. 11

6.4.1. Historical Background

• The church’s present appearance reflects additions and modifications over time, while its original apse remains intact.An inscription indicates that the original wall paintings date to 1171/1172, beneath remnants of 14th-century frescoes.Source: §6.4.1 · p. 11

6.4.2. Digitisation Rationale

• The monument was documented through a collaborative training initiative, with 2023 mission data refined into an HBIM reference model for conservators and architects.The work was part of the EU-funded ERA Chair MNEMOSYNE project and EUreka3D project.Source: §6.4.2 · p. 11

6.4.3. Digitisation Process

• The holistic survey combined TLS and photogrammetry to create an HBIM LOD 400 model and textured interior and exterior meshes.The outputs included georeferenced point-cloud and BIM files, as well as OBJ for online visualisation and STL for 3D printing; digitisation results, metadata, and paradata were aggregated into Europeana through EUreka3D.Source: §6.4.3 · p. 11, Figure 9 · p. 11, Table 4 · p. 12

6.5. Models Processing and Results

• Each digitisation process yielded a 3D model and multiple file formats, enriched with object metadata and paradata about digitisation hardware, software, and conditions.The examples include e57, obj, jpg, stl, and 3dm files.Source: §6.5 · p. 13
• The data-verification workflow checked file correctness, extracted and ingested metadata, and validated metadata and paradata against the EDM scheme.The 3D model files were annotated with the validated information, making the CHOs manageable in Onedata.Source: §6.5 · p. 13
• A dedicated processing workflow converted original OBJ files into multiple GLB derivatives of varying complexity and size for web-based viewing.The workflow containerized the Blender configuration used by the CH team and ran transformations as batch jobs.Source: §6.5 · p. 13
• The next workflow step created public unlisted shares that enabled access to each CHO’s 3D models from the EUreka3D 3D Viewer.The process is described as scalable through Kubernetes deployment and auto-scaling, with workflow executions delegable in parallel across clusters.Source: §6.5 · p. 13
• The published CHO records were registered in Europeana, where live 3D models, metadata, and paradata became publicly available through the embedded 3D Viewer.The example record for the Lambousa Fishing Trawler is described as an EDM record linking metadata, paradata, and the 3D model.Source: §6.5 · p. 13

7. Conclusions

• Detailed 3D digitisation creates standardisation and complexity-management challenges, requiring stronger metadata capabilities for semantic insight into interconnected cultural heritage data.The paper describes metadata as a first-class component of the data management system and notes that its paradata model supports quantitative description and qualitative comparison of artefacts and digitisation processes.Source: §7 · p. 13
• The proposed declarative workflows integrate tightly with data management while allowing execution on different job-processing backends.The authors state that FaaS supports cloud scalability and that workflow libraries and data versioning support reusable workflows and reproducibility.Source: §7 · p. 13
• The solution integrates data, metadata, and paradata management and processing, with adaptable automated workflows and Kubernetes-supported scalability and parallel execution.The authors state that the Europeana publication approach could also be applied to disseminate 3D models through EOSC.Source: §7 · p. 13