DatasetSeries subClassOf Resource instead of Dataset

[jakubklimek]

In DCAT 3, dcat:DatasetSeries is a subclass of dcat:Dataset, which is a subclass of dcat:Resource. In DCAT-AP 3.0.0, it is just a subclass of dcat:Resource, which seems confusing.

[bertvannuffelen]

This issue is related to #278: see https://github.com/SEMICeu/DCAT-AP/issues/278#issuecomment-1710174225 . And also the issue: https://github.com/w3c/dxwg/issues/1387

[terjesyl]

We see the implications listed below if dcat:datasetSeries is just a subclass of dcat:Resource.

We would instead prefer that dcat:DataSeries is a subclass of dcat:Dataset as in DCAT.

• 7.1 Defintion of property ‘type’: Agent is also used as publisher for Dataset Series and should be added.
• 7.2 Definition of class Catalogue should be changed to: “A catalogue or repository that hosts the Resources being described.”
• 7.2 Missing property on Catalogue: There is a property missing in the class Catalogue that links the Catalogue to the DatasetSeries. Without such a direct link, one has two options in order to find all DatasetSeries that are included in the Catalogue:
  • Use the generic property dcat:resource which is not semantically precise.
  • Follow the property dcat:inSeries in all Datasets, which may be a very resource-consuming operation.
  • Possible solution: Add an optional 0..n property datasetSeries as a sub-property of dcat:resource
• 10.2 dct:accrualPeriodicity: Should this also apply to the property dct:accrualPeriodicity in the classes “Dataset member of a Dataset Series” and “Dataset Series”?

[bertvannuffelen]

* 7.2 Missing property on Catalogue: There is a property missing in the class Catalogue that links the Catalogue to the DatasetSeries. Without such a direct link, one has two options in order to find all DatasetSeries that are included in the Catalogue:

This topic was mentioned during the webinar on the alignment with DCAT 3.0. The participants during the meeting did not consider this as a big issue as it could be inferred via querying.

We must highlight that DCAT 3.0 did not introduced this property. And therefore if the need is for such a property, the best place it to amend DCAT 3.0. Otherwise these core organisational properties are in two namespaces and that might lead to confusion.

[terjesyl]

To clarify, this is an issue only if Dataset is just a subclassOf Resource. I might be missing something, but how do you refer from a catalogue to a dataset series if DatasetSeries is just a subclass of Resource?

[matthiaspalmer]

In my view: An application profile cannot and should not try to redefine existing vocabulary with the help of RDFS. Furthermore, adding another rdfs:subclassOf relation between dcat:DatasetSeries and dcat:Resourcedoes nothing as it is already implicit in this case.

Hence, having a statement in the textual document saying that it is a subclass of dcat:Resource rather than dcat:Dataset is misleading. In fact, I think this is nothing more / should not be considered as anything else than a minor bug in the current document.

[bertvannuffelen]

During Webinar 16 jan 2024 the embedding of DCAT-AP classes in DCAT has been explained. There were also 5 principles for writing DCAT-AP application profiles discussed that aid in the further design of DCAT-AP profiles.

[jakubklimek]

Indeed, during the webinar, it was explained that

1. DCAT-AP Dataset is subClassOf DCAT Dataset
2. DCAT DatasetSeries is subClassOf DCAT Dataset
3. but DCAT-AP DatasetSeries is not subClassOf DCAT-AP Dataset

So while I understand HOW this can be done in DCAT-AP, I still cannot find the reasoning behind WHY this is done in DCAT-AP. What are the reasons for not having DCAT-AP DatasetSeries subClassOf DCAT-AP Dataset?

[init-dcat-ap-de]

We would like to understand this as well.

Currently, following the DCAT-AP specification, a dcat:DatasetSeries would not have keywords. (Respectively, it is not easy to understand that they can have keywords.)