- From: Jacco van Ossenbruggen <Jacco.van.Ossenbruggen@cwi.nl>
- Date: Tue, 6 Jun 2017 14:17:52 +0200
- To: public-dxwg-wg@w3.org
On 06/06/2017 01:03 PM, Makx Dekkers wrote: > Jacco, > > Good point that we should look at earlier work. > > I looked through the use cases at https://www.w3.org/TR/dwbp-ucr/ but there are only a few that explicitly address versioning. > > Two requirements were identified: > > https://www.w3.org/TR/dwbp-ucr/#R-VocabVersion: "Vocabularies should include versioning information" > https://www.w3.org/TR/dwbp-ucr/#R-DataVersion: "If different versions of data exist, data versioning should be provided." True for the explicitly identified reqs, but I think the use case section mentions much more stuff that could be relevant. The BBC use case [1] for example, has quite some detail on how you could mint a "versionless" URI (in the format {BASE-URI}/{ONTO-PREFIX}) pointing to the "versioned" URI of the currently most recent version (in the format {BASE-URI}/{ONTO-PREFIX}/{VERSION}), along with the usage of specific metadata properties for current version and prior version, the need to have alternative serialization formats (which can also be seen as a "version" variant), the possibility to time stamp stuff with creation and/or modification dates. The BioRDF use case [2] mentions some issues about versioning, possible in relation to relevant properties from the VOiD/DC/PROV vocabularies. The GS1 Digital use case [3] talks about the expectation management side of things: communicating how quickly dataset consumers should expect new versions to be published, and how likely it is that a static snapshot has been critically outdated by a "live" version. The Land Portal use case [4] talks about versioning in the context of dealing with new versions from other data input suppliers in an dataset aggregation setting, which might also be an issue for e.g. Europeana data sets (Antoine?). > In DXWG, we could look a bit more into the requirements to express the possible *relationships* between versions of datasets which apparently DWBP did not do. There is some high level stuff about this as well, but I agree we might need to look for more specific use cases. Also here there is a lot of work already in popular vocabs including VOiD/DC/PROV/DOAP/OWL. I have not explored to what extend these can be only reused on the solution level, or that they can also be seen as a source of use cases. I would also be interested in knowing _why_ DWBP did not cover this more explicitly. @Ralph, do you remember why even the use of simple predicates to relate versions (such as dcterms:isVersionOf) did not make it to the list of best practices? Jacco [1] https://www.w3.org/TR/dwbp-ucr/#UC-BBC [2] https://www.w3.org/TR/dwbp-ucr/#UC-Bio2RDF [3] https://www.w3.org/TR/dwbp-ucr/#UC-GS1Digital [4] https://www.w3.org/TR/dwbp-ucr/#UC-LandPortal
Received on Tuesday, 6 June 2017 12:18:27 UTC