At all four levels of our data hierarchy, we will aim to ensure MDS meets the requirements of the FAIR principles: for data to be findable, accessible, interoperable and re-usable.
Level 1: institutions
The names of museums (and other institutions sharing data with MDS), linked to information held elsewhere about their location, opening times, etc.
At the most basic level, we will use Wikidata to create new authority records, or enhance existing ones, for all 1,700 Accredited museums in the UK. Wikidata already has records for many hundreds of them, but by no means all. As we compile the level 2 collection summaries described below, we will plug any gaps we find in Wikidata and edit existing records where appropriate.
In particular, we will add alternative names for institutions when we come across them. Museums and their governing bodies can be known by several different names, often the result of past rebranding or reorganisation. Wiltshire Museum, for example, was previously called Wiltshire Heritage Museum and, before that, Devizes Museum. Since older collection records may contain non-current names, it is important that they are captured and associated with current ones in a machine-readable way under a single unique identifier.
As an open source knowledge base, anyone is able to use Wikidata records, and also contribute to them. The Wikidata records for museums often have useful annotations such as geographic coordinates and links to many other online sources. See, for example, the Wikidata record for Wiltshire Museum.
We aim to have created or reviewed a Wikidata record for every Accredited museum by summer 2024.
Level 2: collections
Descriptive summaries of the scope and highlights of collections (and, where appropriate, sub-collections).
Every Accredited museum is required to have a collections development policy that includes, amongst other things, ‘an overview of current collections’. It is good practice for these policies to be made available online, as Wiltshire Museum has done. As is clear from this example, such overviews can pack a lot of useful detail into a few hundred words. However, not all museums publish their policies in this way, and it is not currently possible to search easily across those that are online.
As the second level of the MDS data hierarchy we aim to unlock the enormous research potential of the collection summaries that 1,700 museums have already written by bringing them together as a single, searchable dataset within our repository.
As with the institution-level records already in Wikidata, we are not starting from scratch. We already have collection summaries, dating back to the late 1990s and early 2000s, for around 950 museums that participated in the now-offline Cornucopia initiative. We have reviewed these records, and will be inviting the relevant museums to either approve the existing information, or update it based on their current collection development policy.
We will also contact the 750 museums that were not part of Cornucopia and invite them to share the collection summaries from their collection development policies. We anticipate that most will be happy to do so, but in some cases we may have to create a summary from our own online research.
As with the level 1 records, we aim to have created or reviewed a collection summary record for every Accredited museum by summer 2024 (with sub-collection records as required for larger museums). This will give the most comprehensive overview yet of the nation’s museum collections and be an invaluable research resource in its own right.
Level 3: object records
Item-level catalogue records drawn from museums’ own collections databases.
We estimate there may currently be around 80 million object records just in the databases of the UK’s 1,700 Accredited museums. Most of this data is currently not online and, where it is, it rarely meets the requirements of the FAIR principles to be findable, accessible, interoperable and re-usable. Most seriously, very few UK museums publish their collections data with globally unique and persistent identifiers, the bedrock of FAIR data.
Within five years, MDS aims to bring together – as FAIR data – the object records of at least half the country’s museums, and almost all of them within a decade. If records include the locations of images stored elsewhere online, users should be able to see those images (if they have permission to do so), but we will not be ingesting image files or other digital media due to the prohibitive cost implications.
Museum documentation standards have been around for many years, but there has never been a single data standard strictly followed by all UK museums, nor consistent use of controlled terminologies. (The UK collections management standard, Spectrum, recognises that museums work in many different ways and stresses core principles applicable to various procedures, however these are put into practice.)
So we know the 80 million object records out there will not be consistent, but also know there is enough standardisation, however patchy, for us to work with. We’re going to take these records as we find them. Contributing museums won’t have to scrub their records up into any particular format or to any particular standard. Where needed, we will give records persistent identifiers, and we will map the various field names of incoming datasets to Spectrum’s units of information as a pragmatic head start to the more formal data mappings that some users may need for their specific purposes.
The object records we will hold, therefore, will be the raw material for many different use cases. MDS is not a platform for presenting data from different institutions in a standardised way, as a ‘traditional’ cultural heritage aggregator would do. The harmonising process imposed by such platforms might suit one end purpose, but not others. The nuances and richness of the source records can get lost in translation. In our stripped-back approach the raw data remains available to those who need it that way.
Level 4: new and enhanced data
New content (eg exhibition text) and enhancements (eg AI-generated keywords) linked to level 3 object records.
MDS also aims to be a repository for new and enhanced records arising from the use of the level 3 object records, whether by the museums themselves or in collaboration with third parties.
At a time when museums are under great financial pressure, it seems strange that expensively-produced outputs such as exhibition text or significance reviews are often treated as single-use, quickly gathering digital dust once the project that prompted them has finished. The problem was well expressed by Kevin Donovan back in 1997:
‘Consider for a moment the development of an exhibition and accompanying publication. Labels are written, texts are prepared, all sorts of graphic elements are created … At the end of the day — after tens or hundreds of thousands of dollars have been spent — where is all that content? … The exhibition is now gone … and the content elements created are scattered throughout the organization. Enormous financial and human resources are invested in creating this content, but the results are “one-off”, an unmanaged asset that is largely unavailable for reuse. Imagine the value of accumulating this content over several years and being able to repurpose it on-line.’
Little has changed since then. This is partly because many museums find it hard to manage collections-related information created outside their core systems.
MDS aims to pave the way for an ecosystem of tools and services that will make it easy to create such content with one eye on future re-use. Collections Trust is currently working with several museums to demonstrate the potential of this approach and we will blog about these projects as they progress.