Re: Requirements for a possible "RDF 2.0"

On Jan 14, 2010, at 1:20 PM, Sampo Syreeni wrote:

> On 2010-01-14, Toby Inkster wrote:
>
>> * Explicit support for named graphs
>
> Yes, this would be a step forward. But if you look at them from the  
> logical or the semantic point of view, they are simply another  
> instantiation of reification.

Not really. Names graphs refer to graphs, essentially by ostention,  
i.e. by pointing to them; reification *describes* them.

> If we want to give them a proper axiomatic semantics, we're right  
> back into the debates which surrounded reification from the get go.

No no, there is an easy, clean, semantics for named graphs which  
avoids the reification mess. Its not reification by another name (If  
its done right, anyway :-)

>
> Make no mistake, I think named graphs are a *very* good idea. I just  
> think we might be glossing over a number of minutiae by calling  
> wholesale reification by another name, and just hoping those pesky  
> logicians don't catch upto our newest fad. ;)
>
>> * Literal subjects
>
> Personally I'd like to see subject, predicate and object, all of  
> them, handled on an equal footing. Or if not, I'd like to see an  
> axiomatic semantics which clarifies their inherent difference.

Agreed. There is no reason not to. However, this will make the OWL DL  
folk even less fond of RDF :-).

>
>> * Blank node predicates
>
> Absolutely. Even if nothing else, blank nodes should be allowed in  
> subject, predicate and object. I mean, from my relational database  
> background, I tend to think of blank nodes as the perfect, Coddian,  
> blind surrogate. In that sense, they can serve in any role at all.
>
>> (Though it might be a good idea to phase out blank nodes.)
>
>> From my viewpoint, I tend to disagree. Even in the relational mindset
> they have served well, and RDF's blank node abstraction is perhaps  
> the purest implementation of the surrogate concept thus far. We  
> should not throw it out simply because it makes things a bit more  
> complicated -- instead we should clarify the semantics of it

What is missing is actually syntactic, the notion of scope. If we put  
this back in the RDF syntax, the semantics of bnodes is immediately  
clear. See my ISWC lecture slides for details.

Pat

> , and further utilize stuff like unique properties and other  
> contextual clues to merge various blank nodes into one.
> -- 
> Sampo Syreeni, aka decoy - decoy@iki.fi, http://decoy.iki.fi/front
> +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

Received on Thursday, 14 January 2010 20:38:07 UTC