Re: Naming things with hashes (not #, but e.g. md5)

On 4/11/12 1:46 PM, David Booth wrote:
> On Wed, 2012-04-11 at 13:09 -0400, Kingsley Idehen wrote:
> [ . . . ]
>> Aren't we looking for a clear *definition* of what a URI actually is ?
>>   And then following that, firming up on how a URI can be used to
>> Identify/Name referents that happen to be realm agnostic  *things* or
>> *entities* .
>>
> Yes, that's the part that we're trying to nail down now:
>
>   - An RDF statement author needs to know what conventions to follow when
> writing RDF statements, in order to write RDF statements that RDF
> consumers will be likely to "understand".  (And by "understand" I mean
> that the RDF consumer obtains, a target URI in the statement, the URI
> definition that the RDF statement author intended when writing the
> statement.)
>
>   - An RDF consumer needs to know what conventions to follow if he/she/it
> wishes to discover what URI definition the RDF statement author used
> when writing that RDF statement.

Problem is that RDF has never covered that. On the other hand, Linked 
Data does.

As you can see, conflating RDF and Linked Data has created a unique set 
of mercurial problems. These problems are the fundamental reasons why 
I've always pushed back on point #3 from the more recent edition of 
TimBL's Linked Data meme [1]. The original meme didn't pull RDF and 
SPARQL (which are both implementation details) into the conversation, 
implicitly.

Linked Data is a specific application of the RDF data model that is 
actually representation syntax agnostic. Many don't buy the separation 
of RDF the model and its collection of data representation syntaxes.

>
> Furthermore, if the conventions involve the role of "URI owner" -- and
> not everyone thinks they should -- then:
>
> - The URI owner needs to know what conventions to follow, in order to
> provide a URI definition that is likely to be used by RDF statement
> authors and RDF consumers.

But I really think you are speaking about Linked Data rather than RDF.

RDF rules should be crystal clear in the RDF specs. Linked Data doesn't 
have a spec per se., but TimBL's original meme is a safe 'best practices 
guide' that provides foundation for the kind of standardization I see 
taking shape [2][3].

So back to my point -- excerpted above -- we need to clearly define what 
a URI is what what kinds of realm agnostic *things* or *entities* it can 
name, and how :-)

Links:

1. http://www.w3.org/Submission/2012/02/ -- Linked Data Basic Profile 1.0
2. http://www.w3.org/2012/ldp/charter -- Linked Data Platform (LDP) 
Working Group Charter .
>
>
>


-- 

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