Re: [All] Proposal: RDF Graph Identification

On 08/20/2012 12:04 PM, Antoine Zimmermann wrote:
> Le 20/08/2012 17:14, Sandro Hawke a écrit :
>> [cut]
>> We'd just be separating the RDF semantics from the RDF model and syntax.
>> It would now be okay, according to the spec, to use the RDF model and
>> syntax without using the RDF semantics.
> What does this change to the current situation? RDF semantics is 
> separated from the RDF model & syntax (except that the word "concepts" 
> is used instead of "model") and it is Ok to implement one without the 
> other and this is what people actually do. Not every applications that 
> use RDF have to be reasoners (RDF parsers, W3C RDF validator, RDF 
> editors, some triplestores, linked data browser, etc).

Yeah, it's not clear this proposal would change anything in the real 
world.  It would be a change in how we (some of us, at least) think 
about and talk about the RDF Semantics.    Instead of acting like one 
cannot use RDF without respecting its formal semantics, we would begin 
to accept that as a legitimate option.

My hope is this would allow a more productive discussion about what 
those semantics are good for and why one might want to use them.

My fear is that this will make people ignore the semantics even more, 
starting to do things like make the order of triples significant.   I'm 
not too worried about this, thought, given the momentum of existing 
triple stores.

> Moreover, the current RDF 1.1 Concepts explains much better than RDF 
> 1.0 Concept what the concepts are supposed to mean. So I expect that, 
> with the new specs, it will be even easier to implement the model 
> without bothering about the formal semantics.

Easier-to-implement sounds good, as long as what they are implementing 
is useful and interoperable.

       -- Sandro

> -AZ
>> [cut]

Received on Monday, 20 August 2012 16:16:35 UTC