RE: Webby Data

Hi Jeremy,

 

Thanks for your response. 

 

Maybe I am nit-picking here but I see you talk about URLs, locators for things, while I am more interested in what happens with URIs, identifiers for things. 

 

I can see the value of last-version URLs for document – and I maintained this type of chaining, manually, during my time at DCMI – but I do not see how a durable URL can *identify* something, as the resource it points to can change without notice. It means you cannot use that URI for anything else than pointing to it; you can’t rely on its contents. 

 

But apart from that, in the DCAT perspective, I would not know what the RDF class is of the thing that the ‘durable URI’ identifies. It’s not really a dataset; it’s probably more like a ‘placeholder’ resource that contains a pointer to a ‘real’ resource – just like the ‘latest-version documents’ that I managed were placeholder HTML files with an “include” that imported the real document.

 

How do you see that working?

 

Makx.

 

 

 

 

 

From: Jeremy Tandy [mailto:jeremy.tandy@gmail.com] 
Sent: 12 October 2015 09:01
To: Erik Wilde <dret@berkeley.edu>; Makx Dekkers <mail@makxdekkers.com>; Phil Archer <phila@w3.org>; public-dwbp-comments@w3.org
Subject: Re: Webby Data

 

Hi all. 

Thanks to Phil for pointing out the BP for version identifiers etc. In relation to the possible approach, it says "two URIs both resolve to this document". Whilst this is true, it's worth considering (and pointing out in the doc) that the durable URL (used for the doc throughout its entire lifecycle) points to the resource that is the "document throughout its entire lifecycle and is subject to change" whilst the URLs of versions point to a resource that is immutable; a snapshot in time. 

Makx- you say that it's difficult to apply this to datasets? I think not. The important point is the durable URL always points to what the _publisher_ decides is most appropriate (e.g. the most up to date information set); this resource is mutable. The snapshots are immutable and should be provided different URLs. From my perspective the URLs are point to different resources - in some cases (e.g. the current version URL) they refer to the same information set. 

It's useful to be able to crosswalk between durable and versioned resources. The Linked Data Registry provides a formal description of the necessary relationships and talks about "VersionedThing". 

You can use the properties to express the interrelationships. This would then _imply_ that the resource is versioned. You don't explicitly need to say that it's a VersionedThing. 

That's all for now. Have to travel to work ...

Jeremy 

On Sun, 11 Oct 2015 at 20:45, Makx Dekkers <mail@makxdekkers.com <mailto:mail@makxdekkers.com> > wrote:

Phil,

I agree that a latest version link and timestamps work fine for documents in a publication workflow. Datasets are maybe a little bit more complex. For example, if you add a section to a document, I think that everybody would agree that it's a new version of the document. In the case of a dataset, you could have datasets that change over time but are still considered the same dataset. For example, a dataset recording today's weather may be extended every hour with an additional column, and it still would be today's weather.

Makx.