reducing/reframing role of RDF Semantics; was Re: [All] Proposal: RDF Graph Identification

On 08/16/2012 05:40 PM, Pat Hayes wrote:
> On Aug 16, 2012, at 4:00 PM, David Wood wrote:
>
>> On Aug 16, 2012, at 14:20, Ivan Herman <ivan@w3.org> wrote:
>>>> Mutability only requires changes to the semantics if mutability is perceived as having semantic consequences. So a good place to start might be to ask anyone who thinks it does, to come up with an example involving mutability and entailment (or consistency). For example, if we have explicit time snapshots, then an inconsistency which appears when times can be inferred to overlap might be such an example.
>>>>
>>>>> Pat has said in the past that perhaps we should forgo a semantics for RDF given the widely divergent interpretations in the wild.  I (and others) don't think we can or should just throw away the semantics, if for no other reason than OWL, RIF, etc, currently depend on them.
>>>> Well, actually they don't. Both OWL and RIF have their own, normative, sematnics defined independently of RDF.
>>>>
>>> I cannot check it right now (I write this mail from my mobile) but as far as I remember the RDF compatible semantics of OWL 2 explicitly refers back to the RDF Semantics (of course the 2004 version) in its definiton. To be checked, though.
>> OWL 2004 Semantics clearly references both RDF 2004 Concepts and RDF 2004 Semantics normatively [1].
> Yes, but it also says that the "direct semantics" is normative for OWL and if there is any clash between that and the RDF-compatible semantics, then the direct semantics wins. So OWL can survive even if RDF semantics goes away. In fact, the OWL specs could be rewritten much more simply if RDF had no semantics at all.  I think something similar is true for RIF, but Sandro can speak to that with more authority.

[not without doing some research that seems unnecessary right now]

I'm not sure I understand what you're suggesting.   Let me approach it 
as a strawman proposal which I might support (not sure yet) for 
reframing how we talk about the RDF Semantics, perhaps getting them more 
in line with how they are being really used.

Basically, the idea is we reduce the role of what has been called "RDF 
Semantics" to being another entirely optional "entailment regime" [1], 
like the various OWL and RIF profiles.    Instead of saying this is the 
meaning of the language, we say this is something people might do with 
RDF graphs.

Perhaps the revised document would be titled, "RDF Entailment and 
Consistency", or "RDF Reasoning", or "RDF Logical Operations", possibly 
with the word "Core" in there somewhere.

So, implementers could see it as defining a set of operations 
(entailment checking, consistency checking) which they might want to do 
with RDF Graphs.  Hopefully people not interested in these operations 
would still be encouraged to do things in such a way that others can 
sensibly use them with the same data.    But it would be social and 
market pressures getting people to respect this stuff (as it always 
really is), without a retreat to "because W3C said so".

So -- in a sense, this is not change at all.  But it is a different 
framing.     It's no longer "defining the semantics of the language", 
instead it's defining several entailment regimes to be used with the 
data structure known as RDF triples.

Thinking about this for all of 20 minutes, it sounds sensible to me.

     -- Sandro

[1] http://www.w3.org/TR/sparql11-entailment/

Received on Friday, 17 August 2012 19:30:46 UTC