[v3.2.0] Need to relate Evidence and Output til Dataset

[jimjyang]

The class Evidence is no longer a subclass of Dataset in CCCEV v.2.1.0 (which though is under Public Review right now together with CPSV-AP).

As stated earlier, in order to implement the "once only" principle, we need to relate an Evidence to existing Dataset if any. Conf. also the usage note of the class Evidence which says "Evidence can be any resource - document, ...". Similarly, we need to relate Output to Dataset, conf. also the usage note of the class Evidence which says "the Output of one service will be Evidence for another service.".

What about introducing a new optional property, something like cv:isDescribedAs, with dcat:Dataset as the range, in both Evidence and Output, in order to be able to relate Evidence and Output to Dataset (the "generic" Dataset as defined in W3C/DCAT)?

[EmielPwC]

We agree that it will be useful to relate Evidence and Output to Dataset and to have the relation optional. We will propose such a relation in the upcoming webinar. In order to do so we would like to gauge your desired degree of the relation.
We see 3 main options:

• a relation cv:isDescribedAs which would limit the dataset to only describe one output or one evidence
• to reuse the dct:isPartOf relation as we do for public service
• a completely generic relation dct:relation

Do you have a particular preference and insight on the desired usage? We would recommend to be as meaningful as possible.

[jimjyang]

dct:isPartOf seems to be the best one among the options.

I am unfortunately unable to attend the coming webinar next Tuesday, so here is some more info: Because of the "once only" principle we have to relate the classes Evidence and Output to the class Dataset in our national application profile, in one way or another. We hope of course that the relation we use at the national level, already exists in CPSV-AP at the EU level ("once only" is an EU principle). As for the desired usage, see e.g. #111.