W3C home > Mailing lists > Public > public-ldp-wg@w3.org > December 2012

Re: ldp-ISSUE-9 (Properties in BPR representations): Should properties used in BPR representations be BPRs? [Linked Data Platform core]

From: Henry Story <henry.story@bblfish.net>
Date: Mon, 17 Dec 2012 15:42:18 +0100
Message-Id: <CD40D5A7-14AD-4F85-AC0D-B30FE09E14B9@bblfish.net>
To: Linked Data Platform (LDP) Working Group <public-ldp-wg@w3.org>

On 2 Oct 2012, at 16:12, Linked Data Platform (LDP) Working Group Issue Tracker <sysbot+tracker@w3.org> wrote:

> ldp-ISSUE-9 (Properties in BPR representations): Should properties used in BPR representations be BPRs? [Linked Data Platform core]
> 
> http://www.w3.org/2012/ldp/track/issues/9
> 
> Raised by: Raúl García Castro
> On product: Linked Data Platform core
> 
> "4.1.8 Predicate URIs used in BPR representations should be HTTP URLs. These predicate URIs must identify BPRs whose representations are retrievable. BPR servers should provide an RDF Schema [RDF-SCHEMA] representation of these predicates."

I think the problem is with "These predicate URIs must identify BPRs whose representations are retrievable  BPR servers should provide an RDF Schema [RDF-SCHEMA] representation of these predicates." 

I suggest instead simplifying to:

"Predicate URIs used in BPR representations should be HTTP URLs which when dereferenced provide the definition of said term."

The reference to BPR is indeed to strong - unless one allows that any Web Server that 
just serves up Turtle for a resource is a BPR server ( just one that does not allow edits
or queries, or updates... )

> 
> I understand that predicate URIs SHOULD be HTTP URLs.
> 
> But saying that "These predicate URIs MUST identify BPRs whose representations are retrievable", implies that the properties MUST be BPRs and, therefore, MUST be HTTP resources and, therefore, MUST be identified by HTTP URLs.
> 
> However, before, it is written that "SHOULD be HTTP URLs".
> Am I missing something here?
> 
> 
> I also disagree with the fact that properties are BPRs.
> 
> It is a common practice to reuse properties from vocabularies that you don't own. Therefore, you don't have any control about them (for example, the rdf:type property).

so the question is: is rdf:type defined in a BPR?

> 
> I think that the restriction here could be that "These predicate URIs MAY identify BPRs whose representations are retrievable".

That is too weak. Vocabularies that are not retrievable is an immediate sign that one 
should not use it - either it is out of date, or the people have not yet understood Linked
Data.  A bit like some X who would argue:

 X: I wrote a web site
 Y: where is it? Is it online
 X: No of course it's not online, it's here on my floppy disk"

Nobody would link to such a site, and with good reason.

> 
> And, then, also change the last sentence to: "BPR servers MAY provide an RDF Schema [RDF-SCHEMA] representation of these predicates." (even if we don't mention how, should we?).

yes, here I agree that RDF Schema is the wrong level - why not OWL? 

So I think this question hangs on what a BPR is. Can one argue that 
the foaf vocabulary is a BPR?

Henry

> 
> 
> 
> 

A short message from my sponsors: Vive la France!
Social Web Architect
http://bblfish.net/



Received on Monday, 17 December 2012 14:43:02 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 9 May 2013 13:44:26 UTC