Re: [RIFWG] [Requirements?] A vision for the RIF

> Michael Kifer wrote:
> > The above must be taken in the context of my earlier message
> > http://lists.w3.org/Archives/Public/public-rif-wg/2006Mar/0161.html
> > where I *proved* that the rule set for which those normative rules act as
> > constraints must have some sort of closed world assumption (more precisely,
> > cannot use the normal first-order semantics).
> >   
> Consider for example a national regulation (eg tax regulation,
> or traffic order regulation) and a suprA-national regulation on the same
> subject, as eg the European Commission specifies. Both, the national and
> the supra-national in some cases have to be expressed, and processed
> without CWA. CXonsider eg tax regulations. If tax payers, tax
> declarations, etc, ie the instance level, is not considered, then therfe
> is no need for CWA, a CWA would even not make sense. Now the
> supra-national regulation is a normative specification for the national
> regulation.

Francois,

There are many plausible informal statements, which cease to make sense
once you try to analyze them formally.


> > I did not say that normative rules must be "governed" by CWA, because I
> > don't know what this might mean.
> >
> > If you think that my very short proof has a bug then please point this out.
> >   
> I am not at all thinking there might be a
> nmistake in your proof. I am only thinking that, most ptobably, you have
> been considering (explicitly or implicitley) assumptions that UI am not
> thinking of.

Yes, this is true. I made an assumption, which was a formal definition of a
constraint.

> And I think my example of national and supra-national
> regulations can also be seen as a proof for my argument.

Not at all. Now you have to make a formal definition of a constraint such
that it would make sense in FOL. Such definition has to pass several tests:

1. It should be formal (after all, we agreed that RIF must have a semantics)
2. It should be able to distinguish normal sentences from normative ones.
   (I.e., it should give them asymmetric roles -- otherwise the definition
   is no good.)
3. It should pass the informal test that we should agree that the
   definition is "reasonable". That is, it conforms to our informal
   understanding of what a constraint is.
   For instance, my definition was standard in DBs and LP, so, it passes
   this test.


	--michael

Received on Tuesday, 25 April 2006 13:39:00 UTC