Re: LinkedData != RDF ?

On 5/20/11 1:58 PM, Sandro Hawke wrote:
> On Thu, 2011-05-19 at 20:55 -0400, Kingsley Idehen wrote:
>> Sandro: since I know you seek Linked Data bootstrap too, please
>> encourage TimBL to roll back his Linked Data meme update. Just take
>> it
>> back to how it was, and the delayed bootstrap will happen. Facebook's
>> nearly there, and evolving from their literals based data graphs to
>> URIs
>> and literals based linked data graphs becomes a simple tweak. Ditto
>> Microsoft, Google and others re. their respective data graph
>> representation formats.  We can do this :-)
> I think that would be a short term gain (people using something labeled
> "linked data" and people using URIs to identify things), but a long term
> loss, because (by having different formats for data), still no
> interoperability.

I simply don't agree at all. This isn't about one syntax for data 
interchange that accommodates high semantic fidelity. Its about a broad 
foundation that accommodates multiple syntaxes for expressing triples.

When I read TimBL's meme, here's what went off in my manic open data 
access middleware oriented mind: Aha! We now have a roadmap (with TimBL 
and W3C stamp) for discrete data objects that are inherently platform 
agnostic. I also assumed the actual data object representation (triples) 
would be syntax agnostic with RDF's family of syntaxes on the table 
ready for use.  In short, I had the same response (in my head) re. URI 
abstraction: I saw HTTP scheme as being on the table, but the other 
schemes could play too etc..

There is a pattern of success re. the Web and I decompose it as follows:

1. URIs -- supports multiples schemes with HTTP there to be exploited 
but you can go implement your own schemes
2. HTTP -- platform and data representation agnostic protocol for 
RESTful client-server; use it or pick/make another protocol that better 
serves your needs
3. HTML -- platform agnostic data representation for hypermedia 
documents; use it or pick/make your own markup for hypermedia documents .

Notice, I didn't proceed to XML. More importantly, 1-3 are consistent 
with the wisdom and ingenuity inherent in TimbL's original Linked Data 
meme.
> I wouldn't be terribly opposed to reversing starts 4 and 5, so you do
> triples-with-shared-URIs first,

YES!!!
>   then RDF second, but it seems to me that
> to get interoperability, you need to actually have standards for the
> bits on the wire....    Okay, that came out wrong, because I know you
> know this...    What format are you proposing for serializing these
> triples?

Multiple, and I am hoping the group here can produce a JSON format to 
sit alongside the collection of formats I refer to as the RDF family [1].

Links:

1. http://dbpedia.org/c/KOYATM -- note the footer section that reflects 
my "family" comment: RDF ( N-Triples N3/Turtle JSON XML )

Kingsley
>     -- Sandro
>
>
>


-- 

Regards,

Kingsley Idehen 
President&  CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen

Received on Friday, 20 May 2011 18:57:12 UTC