Connecting media art archives – closer to reality through advances in research data infrastructures

The 2025 Workshop on New Media Art Archiving took place at ZKM | Center for Art and Media Karlsruhe from 5 to 8 February 2025. The special topic of the workshop – Globally Connecting New Media Art Archives – set the stage for the organisation of thematic sessions and expert working groups with over 60 attendees representing a wide variety of research and cultural organisations from across Europe, North and South America, Asia and Australia. 

Rooted mainly in communties and events formed around the ISEA (International Symposium on Electronic Art) annual conference, the impulse for the workshop initiation was driven by seven archives: ACM SIGGRAPH History Archive, Archive of Digital Art (ADA), Ars Electronica, FILE (Electronic Language International Festival), ISEA Symposium Archives, MEMODUCT, and ZKM | Center for Art and Media Karlsruhe. The workshop brought together additional individual experts and institutions from Germany and beyond specialising in the collecting, archiving and preservation of (new) media art, a loose term referring to a wide range of creative work, experimental formats, festivals and performances operating at the intersection of art, science and technology. Dr Lozana Rossenova from the Open Science Lab represented TIB and contributed infrastructural expertise based on ongoing work in the context of NFDI4Culture, Base4NFDI (KGI4NFDI, TS4NFDI), ECCCH and the MediaWiki open source software communities.

Success stories and challenges in the provision of data

The first day of the workshop featured opening presentations from different perspectives, highlighting success stories from small- to medium-organisations dealing with limited resources but striving for opening up data about thousands of events, performances, exhibitions, artworks and artistic networks, media preservation and more. Key challenges relating to data publication, interoperability, and discovery, alongside long-term preservation in the specific context of media art and the great heterogeneity of associated data and data sources were also identified. Besides representatives from the above-mentioned archives, and lightning talks from diverse software and archival projects and initiatives from Cyprus, Italy, Portugal, Slovenia, Spain, among others, a group cooperating on projects using the public Wikimedia platforms (e.g. Wikdiata) and/or MediaWiki software (e.g. Semantic MediaWiki, Wikibase) also presented challenges and best practice examples from the media art field, featuring archival work by AvoinGLAM, LI-MA, Rhizome, Zentrum für Netzkunst, ZKM and more (slides). 

Common ground

The issues highlighted across all opening talks can be grouped around three main objectives for the field:

• Connecting people – many initiatives on the local and international level share many similar challenges, but often work in isolation and attempt to resolve complex issues with limited resources. Establishing channels for streamlined cooperation and know-how exchange would significantly advance the field, and the formalisation of a network or a foundation can set the ground for this, in addition to formalisation of governance structures such as an advisory committee and working groups. 
• Empowering archives – archives with existing systems struggle to keep their infrastructure up to date and meet the demands of the heterogeneous characteristics of media art, while many smaller and/or event-oriented initiatives (e.g. festivals) lack official mandates to establish formal archival infrastructure, yet need to manage large amounts of historically valuable information. Some projects already work with or actively develop open source solutions that can benefit others, but best practice exchange is limited to national or sometimes personal networks. Supporting the documentation and implementation of existing open source solutions, reducing duplication of effort and the need to ‘build from scratch’ can help empower diverse stakeholders and thereby enrich the field as a whole. 
• Connecting archives – even if all media art collections and archives use standardised, open source systems, an overall vision, information architecture and technical infrastructure that can facilitate interconnections across these archives, mutual enrichment and crucially – federated search – is still an important objective identified by all stakeholders present at the workshop. This is where much of the experience from projects dedicated to facilitate large scale data integration, e.g. NFDI, Base4NFDI, EOSC, including citizen-science collaborative-models such as Wikidata, can prove highly beneficial, in order to both avoid past mistakes and the pitfalls of disciplinary silos and to benefit from latest developments across different domains of science. 

Chiara Borgonovo presenting on behalf of the Media Art on Wikimedia projects group and highlighting common challenges in the field. CC-BY 4.0 Lozana Rossenova.

Common architectural vision

Across the three days, three different working groups focused on tackling questions related to what the architecture for connected media archives might look like, what ontology harmonisation work might be necessary, and what end user requirements would need to be met by such a common vision. 

A sketch of the high-level architecture was jointly proposed by Lozana Rossenova (TIB) and Andreas Kohlbecher (ZKM) based on in-depth discussions with representatives from the archival initiatives present at the workshop. 

Diagramme of the overall architecture for connecting media archives not including specifics of the software infrastructure. CC BY 4.0 Lozana Rossenova.

This high level proposal was based on several principles:

• Decentralisation – all archives retain full control of their data, data is not aggregated or duplicated, but connected via a registry hub – itself a knowledge graph (KG), following the model established by the KGI4NFDI service. 
• Flexibility – all archives can retain their existing systems and data models, but will receive support where required to open up APIs, deliver RDF data, or implement a new open source system solution (such as Wikibase, for example) of their choice. 
• Modularity – the architecture is modular, but uses common data exchange standards, so that individual components can be replaced and/or updated when needed, and there is no lock-in to a single software solution.  
• Leveraging latest developments in semantic web and ontology services – federation via contemporary query services (e.g. Qlever) can be highly performant and easier for end-users (via auto-completion features, caching, etc); mapping and harmonisation of ontologies and vocabularies does not need to be a labour intensive manual effort; the API-gateway features of the TS4NFDI service and AI-supported entity linking and deduplication (via services such as Antelope, among others) can support interconnecting archives without sacrificing the idiosyncracity or the detail of the source data. 
• Ethical AI use – AI should be used not to mass crawl and index data via bots that strain server resources on the side of the archive providers, and potentially violate individual copyright specifications applicable to contemporary art, but instead to support automating tedious and labour-intensive processes (e.g. entity linking, deduplication, formulating queries), making the work of already under-resourced institutions more efficient and easier to scale. Open source and custom-trained models (using neuro-symbolic approaches, vector and knowledge graph embeddings) can be used to facilitate natural language interfaces for querying data and lowering learning curve barriers (example of existing application: ORKG Ask).  
• Staying connected to global data hubs – last but not least, the vision for common infrastructure should not create a domain-specific silo for media art, but rather benefit from and contribute to the broader LOD space and global data resources, including Wikidata (and expanded Wikimedia ecosystem), EU Data Spaces (Europeana and more), EOSC nodes, etc. The approach to use a registry based on knowledge graphs (KG) and support federation will support this goal as evidenced in multiple NFDI consortia’s application of KG technologies, and the KGI Base service.

Diagramme of the principles guiding the common architectural vision. CC BY 4.0 Lozana Rossenova.

There are of course various aspects related to handling the specificity of the different archives, the integrity of their unique curatorial viewpoints, individual artist agreements on copyrights, and handling diverse multimedia representations in decentralised workflows, that require further architectural considerations. Preserving accurate provenance especially once federation is used as a means of not only discovery, but also enrichment and data from individual archives is reused in other contexts. Such considerations can be better scoped and defined once the network governance is formalised, project-specific funding is secured and the technical implementation work is underway.

Outlook

To achieve the goals of not only connecting archives via a common infrastructure, but also connecting people and empowering individual organisations to structure their information following best practices, the loose network of archives and organisations present at the workshop in Karlsruhe will organise regular communication channels (a Matrix chat software instance, hosted by ZKM; dedicated monthly online calls with two core focus areas – community management and technology specification); work towards the formalisation of the network into a legal structure with a clear governance model; intensify collaboration through dedicated regional or themaric working groups preparing and submitting funding applications relevant to the different objectives outline above (e.g. Network of COST action funding grants for community work; Horizon Europe or Open Infrastructure grants for the technical implementaion). Research institutions and data infrastructure initiatives in Germany, such as TIB, ZKM, NFDI, can play an important role in supporting these efforts going forward through expertise, open source tooling and collaboration in third-party funded projects. Equally the media art network, its partner archives, data and common infrastructure can contribute significant research data intersecting media arts and sciences developed, produced and/or exhibited in Germany back to NFDI4Culture, NFDI and EOSC nodes, helping to weave the interconnected tapestry of cross-disciplinary work and research driving innovation in socio-technical contexts.

Acknowledgements: Thanks to Felix Mittelberger and Andreas Kohlbecker from the ZKM for the invitation to join and contribute to the workshop. Special thanks to Dragan Espenschied, Susanna Ånäs, Gaby Wijers, and the rest of the Media archives on Wikimedia group for the helpful insights and much needed critical perspective provided throughout the workshop. Thanks to the numerous other workshop participants that contributed to making the workshop an inspiring, diverse and inclusive event.