Re: Issue with top-down and bottom-up semantics

Hi Bob:
  I think that your understanding of variables is consistent with the  
flat interpretation:

Ex (B(x) \/ C(x))

<->

(Ex B(x)) \/ (Ex C(x))

Similarly:

Ax (B(x) /\ C(x) )

<->

(Ax B(x)) /\ (Ax C(x))

A query can be viewed as the constructive refutation of the formula:

¬ Ex Q(x)

Frank

On Oct 26, 2007, at 8:43 AM, Bob MacGregor wrote:

> Hi Francis,
>
> I agree with your objection to the "flat model for variable  
> quantification".  In fact,
> the semantics we use is not flat.  In our query language if a  
> variable appears in two
> disjuncts, but not "outside" of them, it is considered to be two  
> different variables,
> independently quantified. If you flatten, you get a semantics that  
> most users would
> find non-intuitive.
>
> We consider much of  NaF semantics to be a property of individual  
> operators, rather than
> of the language itself (or the model).  Thus, for example, if a  
> user wants to use UNSAID, she
> can, and which uses negation as failure, and if she wants to use  
> NOT (classical negation) she
> can use that.  We don't happen to support two different OPTIONAL  
> operators, but we could if
> there were a need for that.
>
> I'm not objecting to open world semantics per se; there are (a  
> minority of) cases when its
> useful.  I'm objecting to the absence of closed world semantics.
>
> Cheers, Bob
>
>

Received on Friday, 26 October 2007 15:51:25 UTC