By Edward Anderson and Tycho Hofstra

Community standards for data exchange are essential for de-siloing and contextualising data about the Rijksmuseum collection. Linked Art is one such initiative, which was recently used to launch the Van Gogh Worldwide platform. In this blog post we discuss how the Rijksmuseum contributed Linked Art data to this platform and how this relates to the development of our Integration Layer and Data in Context projects.  

Linked Art 

Linked Art is a community-led initiative to make a user-friendly model for composing and sharing data about art. The Rijksmuseum has been part of Linked Art since the beginning. We participate in the weekly Linked Art Editorial Board modelling and exemplar testing calls, so we’ve been involved with the project during the specification development phase. Although the Linked Art model isn’t mature yet (a 1.0 release is expected early this year), we were excited to support Van Gogh Worldwide’s decision to implement it.

Van Gogh Worldwide 

Van Gogh Worldwide launched in November 2020, offering an innovative and high-quality platform that brings together art historical and material-technical information about the work of Vincent van Gogh (1853-1890). The platform currently displays the details of more than 1,000 paintings and works on paper by Vincent van Gogh in the Netherlands, which are consultable by object data, provenance, exhibition and literature data and technical material information. The Rijksmuseum is one of the data-contributing partners of the Van Gogh Worldwide project. 

Our motivation to take part in the Van Gogh Worldwide initiative, as an institution, is twofold: both to widen access to our collection and the metadata which describes it; and also to support an innovative programme which is bringing an artist’s oeuvre together for integrated research and access. We are proud custodians of nine of Vincent van Gogh’s artworks and enthusiastic about sharing our knowledge about them. 

While we are already used to make almost all of our collections metadata available online via our APIs and data dumps, the Van Gogh Worldwide project pioneers in implementing the exciting new Linked Art data standard. What’s been really satisfying over the few months that Van Gogh Worldwide has been materialising, is seeing that theoretical framework manifest as a real, functional interface to museum collections. 

Van Gogh Worldwide and Data in Context 

Linked Art levers the Getty thesauri to standardise the description of concept, person and place entities. So to ship Linked Art data, the most convenient way is to use the Getty Art and Architecture Thesaurus (AAT) for describing materials, techniques and object classes; the Getty Union List of Artist Names (ULAN) for artists; and its Thesaurus of Geographic Names (TGN) for places. But in order to do so, you first need to discover the alignments between your own data and Getty’s. And this isn’t trivial.  

The challenge lies here: For one, domain experts are needed to determine whether descriptions of art historical concepts are the same between thesauri; and also, there are a lot of concepts to map across. For just our own nine Van Gogh works we needed to align 33 different Rijksmuseum terms to AAT. To increase the number of alignments, the Research Services department started the Data in Context project in 2020. Within this project, a combination of automated and manual matching techniques is used to align the bulk of the museum’s most used entities. We’ll discuss the first stage of this project in more detail in a future blog post. 

Linked Art and the Integration Layer 

Participating in Van Gogh Worldwide allowed us to explore two possible roadblocks in implementing Linked Art for the Rijksmuseum. Firstly, there’s risk associated with early adoption: because the Linked Art model isn’t completely mature yet, we wanted to avoid getting into the situation where, in the future, we would have to unpick code delivering data which conforms to a standard which has since shifted. 

Secondly, and more significantly, by implementing Linked Art for a tiny subset of artworks, we would be able to do some of the intellectual groundwork necessary for establishing an alpha version of our own Integration Layer – a new internal platform unifying siloed metadata resources across the museum – which, like Linked Art, is built on top of domain specific models such as the CIDOC-CRM. We created a new ETL (extract, transform, load) process to deliver Linked Art data to enable the Rijksmuseum to contribute its artworks to Van Gogh Worldwide. The challenge was to build this in a sustainable way: writing reusable code, adhering to our own infrastructure principles, and establishing patterns we can later extend for a richer implementation. 

Conclusion 

So, the value of taking part in the data delivery was double: we could increase access to the Van Gogh artworks cared for at the Rijksmuseum; and we could start to develop our Integration Layer which will be instrumental in connecting our own institutional knowledge. By starting with nine artworks, we felt like we were taking exactly the right kind of baby steps we would need to start exploring how to generate the model. You can check out the nine artworks of the Rijksmuseum collection on Van Gogh Worldwide yourself by following this link. It is a taster for how we will, eventually, be able to link together all the data for our own collection of a million objects.