
This is the current version of the specification, including the feedback and comments received during the meeting in June, and an in-depth review performed by the CPSV-AP working group.
Content changes are marked and highlighted. The main changes are:
- Property description of Registry Service (Public Service) is now mandatory to be compatible with CPSV-AP.
- The cardinality of dct:type in Registry Service is 0..n instead of 0..1.
- Corrected namespace for hasChannel: cv:hasChannel instead of cpsv:hasChannel
- Registry Service (PublicSerive) hasChannel instead of url. Also cv:Channel instead of rdfs:Resource. This describes any communication channel (url, phone, etc.) of a public service.
- (Issue #1) Included in the model: Resource, Dataset, Catalog, and DataService.
- (Issue #2) Added a new property isAuthoritative (namespace to be defined) with range rdfs:Literal (xsd:boolean), recommended for resources, datasets, catalogues, and data services.
- (Issue #3) hasCompetentAuthority could be multiple.
- (Issue #4) Eurovoc included as mandatory to describe themes. EU Publications Data Theme Taxonomy NAL is also kept to be compatible with DCAT-AP.
The most relevant change we should discuss is the Output of a Public Service. The model tries to establish the relation between a Public Service (i.e, the Registry Service that manages the base registry) and the main output under this context: the Base Registry (described as a dcat:Catalog). So, according to the CPSV-AP: a cpsv:PublicService -> cpsv:produces -> cv:Output (Output must have title and identifier).
Should we reuse this property to indicate the Output produced by the public service is also a dcat:Catalog? It seems this is compatible. So a base registry (or a registry of registries) would be both a dcat:Catalog and a cv:Output. The required properties to describe them, both in DCAT-AP and CPSV-AP, are also compatible. A new issue will be open regarding this.
Categorisation
Attachment
Shared on

Comments
Thanks for the new version.
I seem to remember that we discussed about to include "data element" to describe a "data set", don't we?
Some of us need to describe the elements that a dataset is composed of, and the " isAuthoritative" is indeed a property of the data element instead of the whole dataset, as we discussed in issue #02