Linking research & learning technologies through standards

Last month, ISO released the long-awaited third edition of the ISO 2146 standard for Registry services for libraries and related organisations. ISO 2146 is a standard of great interest to repository communities, and we have already posted on it at some length, including its use as a basis for the Australian National Data Service’s RIF-CS schema, and the IMS LODE registry model. (The latter post includes a UML diagram of the ISO 2146 classes as of its 2008 draft.) Because of this interest, it is worth describing the standard further.

ISO 2146 started out in 1972 as “Directories of libraries, information and documentation centres”—that is, as a standard for directory listings of libraries. As libraries changed and online access to libraries became critical, the scope of ISO 2146 expanded. This third edition started out in 2001, driven at first by the need to model interlibrary lending services. The standard now is not restricted either to libraries or to loaning services: it is a general information model for all interactions with collections, physical or digital, through well-defined services. Those services still include physical instances like procurement and loans; but they also extent to services online accessing repositories and registries, such as search and harvest.

ISO 2146 has such broad coverage, because it is not a concrete model: it models collections and interactions with them in the abstract. A repository, for example, is modelled as a subclass of collections, associated with particular services. The abstraction also means that ISO 2146 is not bound to a particular machine-readable format, like XML, or a particular set of vocabularies. (The only mandated vocabularies are the ISO standards 4217 for currency, 639-2 for language, 3166-1 for country, and 3166-2 for region.) ISO 2146 models the entities in the collections space, but using it in applications will involve particular profiles of the standard. Those profiles will select formats and vocabularies for its classes, and will restrict its classes to those relevant to the particular context. We have already seen IMS LODE and RIF-CS do this kind of profiling. ISO 2146 also allows for extension of its classes to model entities in more detail; we chose for example to model annotations as a distinct class for IMS LODE.

The basic model of the published ISO 2146 is the same as the drafts used in our earlier work: a single root class of Registry Objects, with common attributes, is specialised into the four main classes which describe how collections are interacted with. Collections are managed by Parties, who can be people or organisations. Parties interact with Collections through Services, which have functions (business processes) and products. Services are set up by Parties as part of Activities. Services in turn have are delivered through Protocols, and are subject to Access Policies; the way they are delivered and the obligations they meet are set out in Service Descriptions. Collections can also be described in detail, including attributes such as coverage, annual increase, and accrual policy. The ISO 2146 standard walks through modelling all these entities, their attributes, and the relations between them. The individual Items present in collections, however, are not modelled: they are out of scope of ISO 2146, and are described in other standards, such as FRBR.

There have been a few changes from the previous drafts. Notably, Enumerations with their Values and Schemes were the way controlled lists were modelled. These have now been called Labels, with Values and Authorities, which generalises the data type to any value associated with a particular registration—such as identifiers.

The standard also includes four informative annexes.

• The General guidelines for building registry services make overall suggestions on how to apply ISO 2146 to describing collections, including service typologies (exemplified by the National Library of Australia’s work with the e-Framework), controlled vocabularies, relation ontologies, usage statistics, and bindings and protocols.

• Case Study 1: Registries of libraries and related organisations walks through building an ISO 2146 model around a library, as a single organisation (Party) running a collection. It includes guidance on how to deal with changes in organisations, such as mergers and dissolution, and illustrates issues around modelling interlibrary loans as a service.
• Case Study 2: Collection service registries illustrates modelling collection service registries. These are registries-of-registries, which list service access points to repositories, and allow users to arrange their own access through service descriptions. We have already discussed how this works with the Global Registries Initiative and IMS LODE. Most of the discussion is on the typology of collections within ISO 2146, which differentiates between Collective Works (such as journal runs or datasets), Repositories, Catalogues (metadata repositories), Collection Sets and Sub-Repositories (e.g. OAI-PMH sets), and Registries (collections of the objects modelled under ISO 2146, for use by registry managers).
• Dublin Core Collections Application Profile, finally, maps ISO 2146 to the DCCAP. The DCCAP is already established as a concrete description of simple collections; DCCAP can be regarded as a binding of the ISO 2146 model, since it is more abstract and generic.

ISO 2146 is a powerful and comprehensive model of collections in the abstract and repositories in the concrete. We have already seen it used sucessfully in modelling, and we believe other projects will benefit from its overview of the space.