W3C home > Mailing lists > Public > public-rdf-wg@w3.org > June 2012

Re: dataset semantics being connected to the state of the web

From: Richard Cyganiak <richard@cyganiak.de>
Date: Fri, 8 Jun 2012 18:51:32 +0200
Cc: public-rdf-wg@w3.org
Message-Id: <2F569156-A937-4EBD-AE02-FE20E8B60656@cyganiak.de>
To: Sandro Hawke <sandro@w3.org>
On 8 Jun 2012, at 14:28, Sandro Hawke wrote:
>> We can postulate the existence of a *specific* dataset, let's call it
>> the “web dataset”, and can say that under the condition above the
>> g-pair <i,G> is true in the web dataset. 
> 
> Yes.     I'm not sure that's the most useful framing, but it's quite
> reasonable.
> 
>> (Formally, this could be done
>> as a semantic extension, let's call it W-entailment (for web). So if A
>> is true then *every* dataset W-entails the g-pair <i,G>.)
> 
> The logicians can correct me, but that seems to me like a non-standard
> way to use entailment.  Whether one statement entails another is
> something that can be determined purely by looking at the two statements
> and understanding the logic of the language they are written in.
> Entailment isn't about what statements happen to be true of the domain
> of discourse.

If it is known a priori that certain statements are true, then even the empty graph entails them.

I was talking about the truth of g-pairs. A g-pair says: “The thing denoted by IRI i has state/content G.” In this hypothetical W-entailment extension, this *is* part of the logic of the language, and it's *not* a statement about the domain of discourse.

It's no different from a logic that assigns, say, truth values to mathematical expressions. In such a language, the truth of a statement like

   "3*7"^^math:Expression math:equals "21"^^math:Expression.

can be taken for granted, no matter the state of the world.

In this hypothetical W-entailment, the state of the web is part of the logic, and not part of the world. This is obviously a giant can of worms, which is why I would formally object to a design that makes this part of the standard semantics of datasets. I'd be ok with it as an extension—it's essentially an interesting academic exercise.

(If you normatively say that the truth of anything depends on dereferencing something, then you'll always end up opening the same can of worms.)

Best,
Richard
Received on Friday, 8 June 2012 16:52:05 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:49 GMT