W3C home > Mailing lists > Public > public-rdf-wg@w3.org > November 2012

Re: RDF-ISSUE-103 (dereferenceable-iris): Make dereferenceable IRIs a SHOULD in RDF Concepts [RDF Concepts]

From: Kingsley Idehen <kidehen@openlinksw.com>
Date: Thu, 01 Nov 2012 13:29:25 -0400
Message-ID: <5092B175.6010302@openlinksw.com>
To: public-rdf-wg@w3.org
On 11/1/12 1:12 PM, Gregg Kellogg wrote:
> On Nov 1, 2012, at 7:17 AM, Kingsley Idehen <kidehen@openlinksw.com> wrote:
>
>> On 11/1/12 10:07 AM, Peter F. Patel-Schneider wrote:
>>> -1 on this from me.
>>>
>>> I have nothing against saying that it is good practice under some conditions, but *SHOULD* is quite a strong thing to say.
>> That's fine, but understand that ultimately this affects how RDF and Linked Data are related.
>>
>> Personally, the association is loose ( via SHOULD) or very loose modulo SHOULD, as you suggest via your -1.
>>
>> We really have to be clear about what the RDF and Linked Data relationship is, sooner rather than later.
>>
>> At the current time, there are a number of folks on this list that still believe (and push) the broken RDF == Linked Data narrative.
> So, it seems we have the following assertions:
>
> 1) All Linked Data is expressed in RDF (from TimBL's design note [1])
No, you can have Linked Data with low entity relationship semantic 
fidelity. TimBL's meme doesn't cover explicit entity relationship 
semantics. You have a single implicit semantic that basically requires 
an HTTP URI that denotes an entity to resolve to useful information 
about its referent.

> 2) Not all RDF is Linked Data
>
> I think this should be addressed in RDF Concepts, and SHOULD seems like the right assertion, as there may be good reasons why IRIs are not de-referencable.

Yes.

>   As a web format, IRIs _should_ be de-referencable, and _should_ result in a representation of the denoted resource. If this fails to be accepted, then I'd suggest that we include the assertion with MAY, but that seems far to weak to me.

It think MAY is what Pat and Peter are suggesting re. their -1's for SHOULD.

>
> Creating a needless separation between "RDF" and "Linked Data" will only weaken both concepts.

The separation is critical for either to blossom. Conflation is 
eternally broken, and ultimately does a disservice to both.

>   We should move towards a view that has them as being just different ways to view the same overriding concept.

No, they are loosely related, but not the same thing. RDF enables you 
produced Linked Data endowed explicit machine and human discernible  
entity relationship semantics. You don't get that with Linked Data in 
the most basic form.

Linked Data is really about webby structured data. Basically, it serves 
as a critical bridge between today's Web of documents and the emerging 
Web of (Linked) Data or Giant Global Graph.

Sometimes it helps to look at the evolution of name resolution:

1. Internet -- machines have names via DNS protocol
2. Web of Documents -- documents have names via HTTP protocol, primarily
3. Web of (Linked) Data -- anything or everything has a name via 
de-referencable URIs where HTTP  is naturally the most cost-effective 
protocol option due to its architecture and ubiquity .

>
> I'm a +1 on Markus' wording.
I am at 0. I appreciate and understand the -1's from Pat and Peter.

Kingsley
>
> Gregg
>
> [1] http://www.w3.org/DesignIssues/LinkedData
>> Kingsley
>>> peter
>>>
>>> On 11/01/2012 09:42 AM, Kingsley Idehen wrote:
>>>> On 11/1/12 9:33 AM, RDF Working Group Issue Tracker wrote:
>>>>> RDF-ISSUE-103 (dereferenceable-iris): Make dereferenceable IRIs a SHOULD in RDF Concepts [RDF Concepts]
>>>>>
>>>>> http://www.w3.org/2011/rdf-wg/track/issues/103
>>>>>
>>>>> Raised by: Markus Lanthaler
>>>>> On product: RDF Concepts
>>>>>
>>>>> Lately there haven been quite some discussions about what formats are valid Linked Data. Everyone agreed that at least RDF is certainly one of them. Nevertheless, nowhere in RDF Concepts there's a normative statement that IRIs SHOULD be dereferenceable which is the core principle of Linked Data. The only statement I found about this is
>>>>>
>>>>> "A good way of communicating the intended referent to the world is to set up the IRI so that it dereferences[WEBARCH] to such a document."
>>>>>
>>>>> I would thus like to propose that a normative statement like the following is added to RDF Concepts:
>>>>>
>>>>> "When deferenced, IRIs SHOULD return an RDF Document that describes the denoted resource by means of RDF statements."
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>> Yes!
>>>>
>>>> Then at the very least, you have much clearer sense of how RDF and Linked Data are related i.e., RDF enables you create Linked Data. Much better than the quantum leap to the distorted realm of RDF and Linked Data isomorphism.
>>>>
>>>
>>>
>>>
>>
>> -- 
>>
>> Regards,
>>
>> Kingsley Idehen	
>> Founder & CEO
>> OpenLink Software
>> Company Web: http://www.openlinksw.com
>> Personal Weblog: http://www.openlinksw.com/blog/~kidehen
>> Twitter/Identi.ca handle: @kidehen
>> Google+ Profile: https://plus.google.com/112399767740508618350/about
>> LinkedIn Profile: http://www.linkedin.com/in/kidehen
>>
>>
>>
>>
>>
>
>
>


-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen







Received on Thursday, 1 November 2012 17:29:52 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:52 GMT