w3c / dxwg

Data Catalog Vocabulary (DCAT)
https://w3c.github.io/dxwg/dcat/
Other
152 stars 47 forks source link

Describing the "Release Type" #1289

Closed riannella closed 3 years ago

riannella commented 3 years ago

Stakeholders

Aviation industry Healthcare industry

Problem statement

The need to describe the "release type" of a dataset, such as "baseline" or "full", "snapshot" and "delta" (etc)

Links

Aviation: http://portal.opengeospatial.org/files/?artifact_id=40502 Healthcare: https://www.digitalhealth.gov.au/newsroom/product-releases/snomed-ct-au-and-australian-medicines-terminology-january-2020-release

Requirements

Related use cases

Potentially related to the Dataset Series requirement

Comments

I am not sure if "release type" is the most appropriate property description (maybe better ways...)


andrea-perego commented 3 years ago

@riannella , thanks for contributing this use case.

The current DCAT3 ED includes two sections describing possibly related aspects:

  1. How to specify the version delta with a textual description: https://w3c.github.io/dxwg/dcat/#version-info
  2. How to specify the "status" of a resource in its life-cycle: https://w3c.github.io/dxwg/dcat/#life-cycle

About point (1), based on the documents you cite, I guess that with "delta" you rather mean a version including only what has been added, or a diff. Should this be the case, your requirement may be related to https://github.com/w3c/dxwg/issues/1271

About point (2), I wonder how much what is in the DCAT3 ED covers your requirements, e.g., in relation to the notion of "snapshot".

Another issue is to understand whether your use case fits in either the notion of version or dataset series.

riannella commented 3 years ago

The "release type" has to do with the temporal aspects related to the dataset. (See this for full details in aviation: http://aixm.aero/sites/aixm.aero/files/imce/AIXM511/aixm_temporality_1.1.pdf)

Perhaps we can use prov:wasGeneratedBy and indicate the activity that lead to this specific release:

prov:wasGeneratedBy http://aixm.aero/temporality/SNAPSHOT

andrea-perego commented 3 years ago

@riannella said:

The "release type" has to do with the temporal aspects related to the dataset. (See this for full details in aviation: http://aixm.aero/sites/aixm.aero/files/imce/AIXM511/aixm_temporality_1.1.pdf)

Perhaps we can use prov:wasGeneratedBy and indicate the activity that lead to this specific release:

prov:wasGeneratedBy http://aixm.aero/temporality/SNAPSHOT

Indeed. This approach is also documented in DCAT 2 - see §C.2 Dataset provenance.

andrea-perego commented 3 years ago

@riannella , I wonder whether this issue needs further discussion.

Should that be the case, could you please share some possible examples of the types of relationships you think could be used to specify the release type?

E.,g., should they just specify the "type" of release (snapshot, delta, etc.), possibly by using a code list? Or there's the need of having (also) relationships linking to different versions / release types?

riannella commented 3 years ago

I've tried to generalise these:

BASE - A release type in which the release dataset contains all permanent properties that are in effect SNAPSHOT - A release type in which the release dataset contain all properties at are in effect at a particular time DELTA - A release type in which the release dataset contains all new permanent properties DELTATEMP - A release type in which the release dataset contains all new temporary properties

I think we just need the types. I don't think there is a need (yet) to show the relationship between them.

andrea-perego commented 3 years ago

Thanks, @riannella .

Based on what you say, I think "release type" is more related to alternative editions of a resource (prov:alternateOf) rather than to the notion of "version" used in PR https://github.com/w3c/dxwg/pull/1295 currently under discussion - see preview here:

https://raw.githack.com/w3c/dxwg/dcat-versioning-v2/dcat/index.html#dataset-versions

About specifying the "types":

What could be used for this purpose in DCAT is dct:type (see §6.4.13 Property: type/genre), together with a SKOS code list for BASE, SNAPSHOT, etc. (to be defined outside DCAT).

Do you think this would work for you?

riannella commented 3 years ago

We planned to use dct:type for something else, but now I think we shall use it for the "release type" (and use our own URIs). Thanks :-)

riannella commented 3 years ago

All done.