Re: SEM: "natural" entailments

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Subject: SEM: "natural" entailments
Date: Tue, 27 Aug 2002 05:24:29 +0100

> Thread: Re: revised version of semantics document
> deep in this thread, Peter wrote:
>  > Why should anyone care at all about the entailment rules?

Actually, this was in response to a message about rules of inference rules.  The
message called them entailment rules, and I didn't change the term.  I was
not arguing at all about which entailments are supported; instead I was
arguing about why it matters whether a rule of inference should have an
existential on the RHS.


> This seems a crucial question.

But now you are going on to a different question.

> If, we insist that the entailments we have are the "natural" entailments 
> then I think we end up with a lot of work. Not least, arguing about what is 
> natural - (we could easily wonder whether God is a web ontologist!).
> 
> My view is that the key role of the entailments is clarification.

Hmm.  Clarification of what?

My view is that the entailments should get at the information that is
implicit in the knowledge.  (That means, technically, that entailment is
defined in terms of inclusion of model sets.)  If that is what you
mean by clarification is something else, then please let me know what you
mean. 

> Yes, of course, we all know that if John is a student and John is an 
> employee then John is both a student and an employee. How does this piece 
> of our knowledge get expressed in our language.
> 
> So far we have two candidates.
> 
> Peter's entailment, in which this "natural" fact appears; and Jeremy/Jos's 
> solipsistic version, in which you have to posit the existence of "both a 
> student and an employee".
> 
> The point about these two candidates is that they are different, and we 
> understand the difference. By the time we are finished we will have chosen 
> one or the other as the entailment supported by Owl. The Owl documents will 
> then line up to show how the entailment or non-entailment holds (or does not).
> 
> After this we will have been clear.
> 
> That's it. That's why this is important.
> 
> Nothing we will say will stop the "natural" entailment from being the one 
> that actually matters in practice to an application, but we can decide how 
> much of that entailment to give to the application on a plate, and how much 
> is outside the scope of what we are doing. The entailments will help us 
> communicate to ourselves and our users the choices that we make.
> 
> I don't see any a priori reason why set theoretic truths have any greater 
> need to be included in Owl than arithmetic truths.
>   Both set thery and arithmetic are really somebody else's concern.
> 
> 
> Jeremy

peter

Received on Tuesday, 27 August 2002 05:46:46 UTC