RE: RIF and QL

> -----Original Message-----
> From: public-rif-wg-request@w3.org
> [mailto:public-rif-wg-request@w3.org]On Behalf Of Enrico Franconi
> Sent: Sunday, January 29, 2006 1:23 PM
> To: gstoil@image.ece.ntua.gr
> Cc: Gerd Wagner; ""Peter F. Patel-Schneider"";
> jos.deroo@agfa.com; public-rif-wg@w3.org
> Subject: Re: RIF and QL
>
>
>
>
> On 29 Jan 2006, at 11:14, <gstoil@image.ece.ntua.gr> wrote:
>
> > Enrico Franconi <franconi@inf.unibz.it> said:
> >
> >> On 28 Jan 2006, at 13:51, Gerd Wagner wrote:
> >>
> >>> { p v q => r. }
> >>>
> >>> and
> >>>
> >>> { p => r. q => r. }
> >>>
> >>>>> are equivalent rulesets, no?
> >>>
> >>>> Well obviously not at least in some readings, as they produce
> >>>> different answers.
> >>>
> >>> Their equivalence (according to the principle called
> >>> "disjunction in the premise") is generally valid in all
> >>> kinds of standard logics if "=>" is read as the implication
> >>> connective. And it also holds in disjunctive logic programs.
> >>>
> >>> However, they may not be equivalent, if "=>" is read as a
> >>> rule operator (not an object language symbol) having the
> >>> epistemic flavor of requiring the condition to be "known"
> >>> (it's not the same to know just p v q or to know p or q).
> >>>
> >>> Since we are assuming standard classical logic (do we?),
> >>> reading rules as plain Horn formulas, Jos is right.
> >>
> >> This is irrelevant at this point.
> >>
> >> In fact, in this case *both* rulesets would not solve correctly the
> >> example in <http://lists.w3.org/Archives/Public/public-rif-wg/
> >> 2006Jan/
> >> 0110>, if a part of 'p' and a part of 'q' are locally evaluated as
> >> queries over some external KB.
> >
> > I don't see why Gerd's clarification is irrelevant. Isn't the case
> > that if
> > you interpret "=>" as an implication then you have the SWRL FOL type
> > semantics and, as you say in http://lists.w3.org/Archives/Public/
> > public-rif-
> > wg/2006Jan/0110, you get what you expected?
>
> Facts:
>
> 1) both rulesets are equivalent if they are given a FOL semantics or
> a disjunctive logic programs semantics.
>
> 2) both rulesets don't solve correctly the example, if the
> interaction with the KB is given by means of local queries over the
> KB, both in the case of FOL and LP semantics for the rules alone.
>
> 3) both rulesets solve correctly the example, if the
> interaction with
> the KB is given by means of global (rules + KB) FOL semantics or of
> global Rosati's LP semantics.
>
> So, I don't see how in this discussion the form of the rules plays a
> role. It is a matter of the semantics of the interoperation between
> rules and KB.

Oh! I see. But in your use case you state that: "In a framework with a
classical semantics we actually get ..., as expected", while other
approaches integrating rules with ontologies might fail, since they operate
in a preferred minimal model. So as far as I understand SWRL belongs to the
classical semantics framework (if not then which approach does), while
Rosati's approach belongs to those of the second category which manage to
get what expected.

Greetings,
G. Stoilos

>
> cheers
> --e.
>

Received on Tuesday, 31 January 2006 08:57:11 UTC