Re: A basis for convergence and closure?

On 2002-02-06 19:53, "ext Pat Hayes" <phayes@ai.uwf.edu> wrote:


> I think I can summarize a way to have all the above with a minimal
> imposition of particular idioms.

I think what you propose is 99% what we need,
with only one glitch, but one which we can
overcome easily enough by simple omission.

> _:s xsd:realnumber "10.3" .
> _:s ex:germannumber "10,3" .

Idioms of this form still will not work in
conjunction with global range constraints.

How do you reconcile the above two triples
with

   age rdfs:range xsd:realnumber .

Even if we know that the value space of
ex:germannumber intersects with that of
xsd:realnumber, the lexical spaces and
mappings are not the same.

Also, what is xsd:realnumber? Is it a datatype
or a property? Both?

And how does the agreed treatment of multiple
range constraints as an intersection apply
to multiple subclassed rdf:value properties?

How do we exclude

  age rdfs:range xsd:realnumber .
  age rdfs:range ex:germannumber .

because the lexical spaces of these types
are disjunct, but not exclude

 _:s xsd:realnumber "10.3" .
 _:s ex:germannumber "10,3" .

???

Do we restrict range constraints to only apply
to literals bound by rdf:type alone? How then
do we actually define constraints/expectations
on local typing?

Finally, I think we will be losing alot
of "brownie points" with the RDF community
by offering two local idioms, one of which
requires all kinds of subclassing mechanisms,
etc. rather than just one, which is clearly
and intuitively synonymous with the global
idiom.

So, I humbly but sincerely ask that we drop
the S-A idiom. Please. Is it really so useful
and special to be worth all the extra effort
and machinery to make it work?


> Oh, I forgot; it also follows that (Dan C., are you reading this?)
> 
> VIII. The in-line use of literals, as in
> <mary> <age> "10" .
> has a fixed meaning which is absolutely unchangeable, which is that
> Mary's age is the *actual literal*, ie the character string '10'.  So
> if you want to write things like that and have them mean Mary is ten,
> then either <age> has to have an odd extension, or else, tough. This
> is where Sergey's idea about XML styles might be a point worth
> making, however, to keep out the noisy townfolk; and as Patrick says,
> you can always interpret <age> to *mean*
> (lambda (x y)
>   x aged _:s
>   _:s rdf:value y  )
> and then put range constraints on <aged>.

Cool. And for the sake of backwards compatability as well
as for those who want to pretend that literals always mean
the same thing, we can simply leave off any default transformation
from 'inline' to bNode idiom, requiring folks that want datatyping
to use the bNode idiom from the start, or do the convertion
themselves. RDF remains neutral on the subject.

The serialized forms for the bNode global idiom aren't all
that cumbersome really:

   Bob age [ rdf:value "30" ] .

   <rdf:Description rdf:ID="Bob">
      <age rdf:value="30"/>
   </rdf:Description>

and since N3 has so much more freedom to change than RDF/XML,
one could concieve of some syntactic sugar such as

   Bob age !"30" .

or whatever floats folks boats.

Cheers,

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Wednesday, 6 February 2002 13:38:38 UTC