Conference Agenda

Heritage science research relies on precise documentation of where samples are taken from on an object and on maintaining clear links between sampling locations, the resulting samples, analytical processes, and interpretations. While IIIF is widely used for image delivery and visual comparison, its potential for linking precise image-based locations directly to structured research documentation is still emerging.

This lightning talk presents work undertaken by the National Gallery within the UKRI-funded RICHeS Heritage Science Data Service (HSDS) project, in collaboration with Kartography (the developers of ResearchSpace). The work focuses on the development of a digital sampling point system for paintings documentation that uses IIIF annotations to connect sampling locations on high-resolution images with Linked Data–based research records.

Building on an earlier bespoke prototype, which is being migrated into ResearchSpace to natively support IIIF enabled, sustainable, extensible, and interoperable documentation workflows. By embedding sampling points within a standards-based knowledge platform, IIIF annotations act as anchors linking images to samples, materials, analytical events, protocols, and interpretations. The approach demonstrates how IIIF can play a central role not only in image presentation, but in integrating images, metadata, and semantic research data to support transparent and reproducible heritage science workflows.

This lightning talk presents an open-source system enabling conversational exploration of IIIF manifests through a Mirador 4 plugin. The solution combines vector embeddings, semantic search, and Large Language Models to make digitized collection content intellectually accessible through natural language interaction.

The system comprises a FastAPI backend that processes IIIF annotations into vector embeddings using local Ollama (stored in PostgreSQL with pgvector), and a Mirador plugin providing an interactive chat interface with real-time streaming responses. Users ask natural language questions about manifest content and receive evidence-based answers.

The demonstration will showcase the complete chat workflow: asking questions, viewing streaming responses, examining relevance-scored evidence. The architecture prioritizes transparency and institutional control—embeddings are generated locally for data sovereignty, while every answer includes verifiable sources addressing AI hallucination concerns.

CODICUM (codicum.eu) is a multi-disciplinary research project about medieval manuscript fragments in Northern Europe. Major goals are to reconstruct lost books and chart the flow of ideas through them. As projects and their funding come and go, many research infrastructures follow that exact pattern. Once a project ends the clock usually starts for when users see tools and services break down or disappear altogether. For CODICUM the decentralization aspect of IIIF is not a nice-to-have feature, it's a necessity.

In this lightning talk we introduce you to the principles that we hope will ensure the sustainability of the technical outcomes from CODICUM so that future researches and users can benefit far beyond the CODICUM project. Our hope is that this talk will lead to others sharing success and failure stories and for us and others to walk away from the IIIF Annual Conference with new ideas and approaches to planning for sustainability.

The CommonsDB initiative has developed a decentralized, trustworthy registry for public domain and openly licensed digital assets. The goal is to empower users to confidently reuse digital cultural content by providing reliable information about its copyright status and attribution requirements. The registry enables verifying content to retrieve rights declarations whereby content providers assert the status of assets. It recognizes content that is similar to the content it has received declarations for, and can flag conflicts if various parties have declared different rights for a work. CommonsDB uses International Standard Content Codes (ISCC) to implement content-based search, and Verifiable Credentials to warrant trust in the declaring parties.

This presentation will report on the project’s results - by June 2026, the registry will contain over 5M declarations.

A registry like CommonsDB can host declarations for IIIF resources. For IIIF users this could be a valuable complement to the metadata in IIIF manifests. We will investigate how this could be done and report on it. We will also assess the relevance of having the CommonsDB declaration workflow tuned for IIIF resources. Either could help position the CommonsDB registry in an ecosystem of “IIIF Content Commons” providing cross-institution discovery services for IIIF resources.

IIIF-in-a-Box is a lightweight, configurable platform. It is designed to make it easy for institutions to deliver IIIF-based digital collections, exhibitions, and computational access environments. IIIF-in-a-Box packages the tools, components, and workflows needed to deploy a self-contained IIIF environment quickly and reproducibly.

Exhibitions and research projects may require temporary, self-governed spaces. These spaces need dedicated computational resources and clear separation from institutional systems. IIIF-in-a-Box supports this by providing a project-based environment with its own governance model. It includes isolated OCR content and the ability to embed front-end components into existing CMS pages. At the same time, it enables institutions to closely monitor resource usage.

A major focus of IIIF-in-a-Box is enabling computational access to collections. Data is exposed in standard, machine-readable formats. These formats are suitable for AI workflows and for emerging research users who want to compute over collections.

IIIF-in-a-Box allows institutions to control access at the level of each “box.” This supports responsible experimentation while limiting attack surfaces. It also helps institutions respond to increasing pressures, such as bot-driven harvesting of cultural heritage data.

The Source Library (sourcelibrary.org) is an open digital library on a mission to translate the Renaissance. Born from the UNESCO-recognized Bibliotheca Philosophica Hermetica (Embassy of the Free Mind) in Amsterdam, it curates primarily 15th–18th century texts in Latin, German, Hebrew, Arabic, and more. IIIF and compatible image APIs serve as a universal input layer, ingesting books from 13+ institutional providers including Gallica, MDZ, Bodleian, Vatican, e-rara, Cambridge, and the Internet Archive. Each book's page images flow through Gemini vision OCR, automated translation, illustration detection, and DOI-minted scholarly edition publishing — with the same downstream pipeline regardless of the source institution. The result: over 2,000 books identified as first English translations and nearly 70,000 extracted illustrations. IIIF's interoperability promise is transformative when the consumer is an AI model.

IIIF manifests already link to Creative Commons and RightsStatements.org for usage rights - but what about preservation intentions? When institutions migrate formats, make copies, or transfer collections, whose guidance shapes those decisions?

We seek to complicate the interplay between property, privacy, and perpetuity (but in a good way):

Privacy has a different timeline to copyright.

Cultural protocols are more complex than property rights.

Machine learning needs context, not control.

Our lightning talk introduces KeepRight, a protocol for expressing preservation intentions that complement existing rights frameworks. We'll demonstrate how KeepRight integrates with IIIF manifests and invite institutional partners for our June 2026 v0.1 launch, and we'll have a poster too, so come and tell us what you think!