CCSI Digital Twins

The research problem

Cultural and creative assets rarely exist as isolated digital objects. A musical work may be expressed through scores, lyrics and arrangements, performed by different artists, fixed in sound or audiovisual recordings, distributed through different releases and files, and documented by several organisations. Films, books, photographs, games and heritage objects have similarly complex relationships between intellectual creations, physical or digital manifestations, documentary evidence, people, organisations, rights and conditions of use.

Conventional metadata records usually capture only particular views of these relationships. A library catalogue, museum collection system, rights registry, distributor database or streaming service may each describe the same asset differently because each serves a different institutional purpose. Our research investigates whether Digital Twins can provide an operational knowledge layer connecting these different forms of knowledge without requiring authoritative systems to be replaced or collapsed into a single database.

This research brings together work on semantic modelling and Linked Open Data, evidence and knowledge production, rights information, human review and interoperable knowledge infrastructures. An important foundation is the work of the Semantic Computing Research Group (SeCo) at the University of Helsinki, including its research on semantic harmonisation, the Sampo model, the 8-star Linked Open Data model and knowledge-graph applications.

From Heritage Digital Twins to Creative Asset Digital Twins

The Heritage Digital Twin Ontology (HDTO) developed in the context of ECHOES provides an important starting point. HDTO provides a semantic framework for representing Heritage Digital Twins and connecting information concerning heritage assets, their components, representations and associated knowledge. Our work with the Dēliņi Farmstead provides a practical environment in which these concepts can be exercised against heterogeneous museum, archival and photographic evidence. :contentReferenceoaicite:0

The qualification heritage, however, describes only one possible status of a cultural or creative asset. A recently released sound recording, film, photograph, book or other copyright-based creative asset does not need to acquire heritage status before the same problems of identity, representation, provenance, interoperability and knowledge maintenance arise. The distinction can become particularly artificial where the same asset simultaneously participates in commercial, archival, research and heritage contexts.

Building on a research direction proposed by Jouni Tuominen and the Semantic Computing Research Group, we aim to investigate an extension of HDTO towards copyright-based creative assets: a Creative Asset Digital Twin Ontology. The objective is not to replace HDTO, but to extend its applicability while maintaining semantic interoperability with HDTO and other established and emerging cultural heritage data models. The research includes analysis and benchmarking of creative-asset datasets and existing knowledge-representation approaches, together with formal validation mechanisms such as SHACL.

Works, manifestations and rights

Extending Digital Twins to copyright-based creative assets introduces requirements that become particularly visible in music and audiovisual domains. An intellectual or artistic work is not identical to the manifestations through which it is distributed and experienced. A musical composition can be performed by different musicians and fixed in different sound recordings; a film can exist through different releases, restorations and digital representations; a literary work can have different editions and translations.

These distinctions matter operationally because different layers may have different creators, contributors, identifiers, provenance and rights. A work entering the public domain does not imply that every manifestation associated with it is free of rights. A contemporary sound recording of a public-domain composition, for example, may remain protected through performers’ and phonogram producers’ rights. Similar distinctions arise for audiovisual recordings, translations, editions and other manifestations.

Creative Asset Digital Twins therefore need to represent not only descriptive characteristics of cultural objects but relationships between works, performances, manifestations, digital representations, people, organisations and applicable rights. This makes the extension particularly relevant to Asset Reactivation, where establishing what can actually be accessed, licensed, published or reused is often as important as identifying the underlying work.

Semantic modelling and validation

SeCo brings extensive experience in harmonising heterogeneous cultural datasets into shared semantic representations. Its reproducible data-transformation workflows combine approaches including metadata conversion, named entity recognition and linking, entity reconciliation, semantic enrichment and knowledge-graph generation across cultural and societal domains.

This provides an important methodological foundation for the Creative Asset Digital Twin research. Rather than requiring source organisations to adopt a single internal information system, semantic modelling can provide interoperable representations through which knowledge originating in heterogeneous systems can be connected and reused.

Formal validation is an important part of this architecture. The research builds on SeCo’s 8-star Linked Open Data model, which extends the conventional 5-star deployment scheme by emphasising explicit machine-processable schemas, formal data-quality validation and explanations supporting assessment of whether data can be trusted as a representation of the real world. We aim to apply and further test these principles with copyright-based creative assets and develop practical approaches for creating, validating and publishing interoperable linked data.

Digital Twins as operational knowledge

A Digital Twin is more than a static metadata record. Knowledge about a cultural or creative asset can change as new documentary evidence appears, entities are reconciled, attribution changes, new manifestations are created, rights expire or are transferred, and institutional or legal circumstances change.

Our work on the Dēliņi Farmstead provides an empirical environment for investigating these operational questions in a heritage setting. Starting from heterogeneous museum records, archival descriptions, photographs and other documentary evidence, the work examines how entities and relationships required by HDTO can be reconstructed and maintained in an operational knowledge infrastructure. In particular, it exposes practical questions around persistent identity, granularity and mereology, digital representations, provenance and the relationship between source evidence and semantic assertions. :contentReferenceoaicite:1

The Dēliņi case and copyright-intensive creative-sector use cases play complementary roles in this research. Heritage collections provide demanding examples of heterogeneous historical evidence and complex physical objects, while music, film and other copyright-based assets foreground relationships between works, manifestations, contributors and rights. Together they provide empirical requirements against which more general Creative Asset Digital Twin models can be developed and tested.

The Sampo model and application layer

The Sampo model, developed by SeCo, provides an established architecture for publishing and using Linked Open Data. Among its principles are collaborative data creation and publishing, shared ontology infrastructures and a clear separation between the Linked Open Data service and the applications through which users interact with it.

This separation is particularly useful for Creative Asset Digital Twins. The semantic knowledge representation should not be identified with a particular user interface or application. Different applications can provide different views and operations over the same interoperable knowledge infrastructure, while the underlying identities, relationships and provenance remain available independently of those interfaces.

Sampo-UI, SeCo’s open-source framework for building knowledge-graph applications, provides one potential application environment for this research. An important research direction is extending knowledge-graph interfaces beyond search, exploration and visualisation towards human inspection, correction and review of individual records. Filtering and visualisation can also contribute to identifying possible biases, inconsistencies and quality problems in source data and AI-assisted knowledge production.

Human-centred and neurosymbolic knowledge systems

Creative Asset Digital Twins also provide a meeting point between symbolic knowledge representation and contemporary AI. A promising research direction is neurosymbolic AI, in which generative AI and large language models operate together with curated knowledge graphs rather than treating either paradigm as sufficient on its own.

Knowledge graphs provide explicit identities, semantic relationships, schemas and validated knowledge that can constrain or contextualise AI-assisted operations. AI methods can in turn support extraction, enrichment, reconciliation and interaction with complex knowledge structures. This connects Digital Twin research directly with our work on Evidence & Intermediate Representations and Human-Centred Knowledge Production.

The objective is not to allow generative models to overwrite curated knowledge. AI-assisted processes can contribute observations, candidate assertions and suggested relationships, while semantic validation, provenance and human review provide mechanisms for determining what becomes part of maintained operational knowledge.

Interoperability without system replacement

Creative Asset Digital Twins are not intended to replace the specialised systems already used by museums, archives, libraries, collecting societies, publishers, producers, distributors and platforms. Those systems embody different institutional responsibilities and frequently remain authoritative for particular kinds of information.

The research objective is instead to establish interoperable knowledge representations through which relevant assertions can be identified, reconciled and reused while retaining their provenance and institutional authority. A rights registry can remain a rights registry, a museum catalogue can remain a museum catalogue, and a distributor can retain its operational database while selected knowledge becomes available through shared semantic representations.

This separation also allows the Digital Twin itself to remain technologically open. Persistent identities and semantic relationships need not depend on a particular database product, knowledge-graph implementation or user interface, and applications can consume appropriate projections of the available knowledge without becoming authoritative sources themselves.

Research & implementation

Our research brings together several complementary strands: extending Heritage Digital Twin concepts towards copyright-based creative assets; representing works, manifestations and rights; semantic modelling and formal validation; evidence-based and human-centred knowledge production; the Sampo model and knowledge-graph interfaces; and neurosymbolic approaches connecting curated knowledge with generative AI.

The Dēliņi Farmstead provides a heritage demonstrator for testing Digital Twin concepts against heterogeneous real-world evidence, while music, audiovisual and other creative-sector use cases expose requirements associated with manifestations, rights, licensing and asset reactivation. These empirical environments allow the research to proceed from concrete interoperability and operational problems rather than attempting to define a universal Creative Asset Digital Twin model in advance.

The longer-term objective is to develop Creative Asset Digital Twins as interoperable, rights-aware and maintainable knowledge representations for cultural and creative assets throughout their commercial, archival and cultural lives, while retaining compatibility with the emerging European Heritage Digital Twin ecosystem.