Call for Comments - Linked Open Data Recommendation for Museums

CIDOC recognises the accelerating trend for cultural resources to move to digital environments. Huge digital libraries and aggregation services are collecting data and metadata about cultural objects and in particular museum objects. Examples include the Europeana Digital Library, the Finnish "CultureSampo" and the German Digital Library. New technologies such as RDF Triple Stores have recently become available with the necessary scalability for managing these large datasets. This revolutionises the approach to description of resources, moving from local documents to semantic networks of billions of elementary global statements that can relate anything in the world. The link between these statements and the resources they describe are made with "Universal Resource Identifiers" (URIs). These are character sequences that should be assigned to only one thing in the world, like the serial numbers of cars.

In addition to their use in large triple stores, URIs also benefit the sharing of information about local collections of any size on the Internet. By relying on unique identifiers (URIs) authorized by the curators of museum objects, other actors can provide supplementary, enriching data to the original datasets. URIs are also provided by and needed in various authority services providing standardized name forms, thesauri or lists of identifiers. URIs are, furthermore, also used to facilitate different kinds of automated, scholarly or social annotations in order to enrich our knowledge about cultural heritage resources.

In the emerging, global information infrastructure for cultural heritage, information providers must create a URI for everything they wish to describe. However, a unique identifier is still not useful without an explanation of its meaning. The Linked Open Data (LoD) initiative (see link below) aims at providing an authorized descriptive dataset to support the unambiguous identification of objects. This method is being rapidly adopted by many organisations, including OCLC (www.viaf.org).

There is, however, an emerging problem. LoD and the Semantic Web have no declared policy about who will create a URI for well-known things like Shakespeare or the Acropolis. Anybody can do it. Now, while these particular examples are external to museums, many museums are delivering huge amounts of content to digital aggregation services. These services are in turn creating URIs for every piece of information they get from the museums and every aggregator creates their own URI even if it is for the same thing. So the billions of things we have are each getting multiple identifiers, each from a different agency and each isolated from all the other identifiers for the same thing. As a consequence, to connect to all the information about one object we have to find which other identifiers in the sea of tens of billions of identifiers relate to the object we are interested in. Most of the aggregators are not in a position to verify, or ground-truth, their identifiers, because they have no access to the objects themselves. However, the museums, as a basic part of their scientific and legal mission, preserve the identity of an object within their inventory and so can produce identifiers that are automatically ground-truthed.

So, if museums do not rapidly deal with the approach that Semantic Web aggregators are using to identify their objects, a quagmire of competing authorized and unauthorized content about museum holdings will develop. The scale of the problem will be of such a magnitude that nobody will ever be able to unpick the mess. We therefore regard it as mandatory, that museum objects are identified in such a way that verification of the identifiers against the real world objects is a simple and effective process. Furthermore it should be simple to curate these identifiers.

Therefore the CIDOC CRM SIG and the Co-Reference Working Group propose that CIDOC adopts the attached recommendation. It is an extremely simple and easy-to-understand mechanism that, in its minimum form, costs a museum no more than a line of text on a Website or in a catalogue. The British Museum has already adopted the recommended policy, and the Europeana expert group, Representatives of the European ATHENA Project, and Emmanuelle Berm~{((~}s, chair of the W3C Library Linked Data Incubator Group have given positive responses to the resolution text.

We hereby call for comments from CIDOC and other ICOM members to further prepare this recommendation for formal adoption by May 30, 2011.

Nick Crofts, chair, CIDOC-ICOM
Martin Doerr, chair, CIDOC CRM-SIG
Mika Nyman, chair, Co-reference Working Group

Resolution text: English, French, Spanish