Re: new semantics initiative

On 2002-06-14 19:32, "ext patrick hayes" <phayes@ai.uwf.edu> wrote:


>> (and no fair peeking into URIs either, those are opaque, no?)
> 
> They are referentially opaque, but if Im a software agent I can
> certainly look at them. They are just character strings, after all.
> And I can check whether or not they are in some list I might have of
> reserved character strings. That list, of course, came from God: I
> was born with it inside me. My creator just knew about those strings.
> I guess he had read some W3C spec or other.

OK. If by "RDF vocabulary" we simply mean the list of URIrefs which
denote terms of that vocabulary, and it is simply a list, and the
list itself need not have itself any URIref denotation, then I don't
have any problem with that.

*If* that list of URIrefs does need its own URIref to identify it,
then I simply want to make it brutally clear to users that it need
not correspond to *any* XML Namespace prefix URIref.

The fact that the XML Namespace prefix URIref for the RDF Vocabulary
(list of term URIrefs) is also taken to denote that list of terms
is simply a coincidence, and not based on any quality or constraint
in inherent in XML Namespaces.

That's my point.

>>>> 
>>>>>   (6) Does this require any changes to syntax/ test cases/ Ntriples/
>>>>>   datatyping/ whatever?
>>>>>   A: No.
>>>> 
>>>>  I don't see how it would not. We would need a mechanism in RDF/XML
>>>>  for setting the dark bit on statements and also an explicit
>>>>  representation of that bit in NTriples (such as ';' rather than '.').
>>> 
>>>  No, that is not the proposal.
>> 
>> That's *my* proposal.
> 
> OK, I vote against that for the reasons outlined in the message that
> Guha and I sent about nonmonotonicity.

I don't see how that introduces nonmonotonicity any more so than
saying that some predicate is "dark". Whether the mechanism is
the predicate URIref being a member of some "dark" vocabulary,
or a vocabulary neutral syntactic marker, both mark the triple
as dark. I don't see the difference.

>>>> 
>>>>  But that probably is not a great amount of work, and likely
>>>>  could be done in a backward compatable manner.
>>>> 
>>>>  [In case it's not clear, I'm pretty much in favor of providing for
>>>>   these layering tweaks to the MT and elsewhere, so long as they
>>>>   are not based on reference to namespaces]
>>> 
>>>  I do not follow your reasons for objecting to the idea of a set of
>>>  URIs having an owner. Isnt that a given, in all these discussions?
>> 
>> You've misunderstood me. I have no problems with a set of URIs having
>> a single owner.
> 
> OK, then, that is all we need to make this work. Only the W3C has the
> authority to reserve a vocabulary that the W3C owns.

Sure.

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 Sunday, 16 June 2002 07:25:50 UTC