Minting URIs for the Open Data Service

Everything that the Open Data Service knows about is identified with a Uniform Resource Identifier (URI).

Where these don’t already exist, we need to specify schemes for creating them.

General principles

For data controlled at the institutional level, we should use URIs in the https://data.ox.ac.uk/id/ namespace. In the simple case, URIs should take the form https://data.ox.ac.uk/id/[type]/[identifier].

Where possible, the identifier should be canonical and controlled by an institutional data provider. This identifier should be applicable to all instances of a particular thing in the University. For example, we wouldn’t use an SSO username to identify people, because not everyone relevant to the University has one.

Where there are competing identifier schemes for a particular type of thing:

• if the identifier schemes are used for disjoint subsets, then we can use them, but encode the identifier scheme in the URI. For example, courses managed by Daisy have a daisy component in their URI.
• if multiple schemes can apply to a given individual, it may be necessary to create a new identifier scheme for use in the URI.

The [type] path component should be a slugified string that identifies the class of things being described. However:

• steer away from type names that won’t necessarily be constant for the life of a thing. For example pph (Permanent Private Hall) would be bad as PPHs sometimes get “promoted” to colleges (e.g. St Edmund Hall).

• Where types can be split into sub-partitions, consider using more-specific type identifiers. For example, course would be vague if:

  1. you plan to describe both undergraduate and skills courses
  2. those courses are managed entirely separately, and would never be managed using the same system.

  In this case, use something like degree-programme and skills-course.

Existing URI namespaces

This is an incomplete list of URI namespaces are already in use within the Open Data Service.

https://data.ox.ac.uk/id/graph/[path]

RDF graphs (of type sd:Graph). [path] is usually structured using the source name, and then another component to signify a partition or whether it’s a dataset, or dataset metadata. Examples of [path] include oxpoints/data, oxpoints/metadata, vacancies/archive and vacancies/current. A list of all current graph URIs in use can be seen at https://backstage.data.ox.ac.uk/stores/public/data/.

https://data.ox.ac.uk/id/dataset/[path]

Datasets, VoID or otherwise. Each should be a dcat:Dataset instance. Generally [path] will be a single component, but where datasets are composed of sub-datasets, it’s generally a good idea to create a hierarchy. Examples include catalogue, oxpoints, courses, and courses/medsci.

http://oxpoints.oucs.ox.ac.uk/id/[id]

Entities in OxPoints, e.g. departments, buildings, committees. IDs are arbitrarily-assigned eight-digit strings.

https://data.ox.ac.uk/id/itservices/service/[id]

Services offered by IT Services. IDs are SharePoint row IDs from the Service Catalogue list.

https://data.ox.ac.uk/id/qb-data-structure/[id]

Data structures for the Data Cube vocabulary. Each should be a qb:DataStructureDefinition, with sub-resources (measures, slices, etc) using URIs below that of the data structure definition resource.