Re: MISC: Internet Media Type registration: proposed TAG finding

>On Wed, 2002-05-29 at 12:13, patrick hayes wrote:
>>  >On Fri, 2002-05-24 at 19:09, Peter F. Patel-Schneider wrote:
>>  >>  From: Dan Connolly <connolly@w3.org>
>>  >>  Subject: Re: MISC: Internet Media Type registration: proposed 
>>TAG finding
>>  >>  Date: 24 May 2002 16:13:32 -0500
>>  >[...]
>>  >>  > The author of the best-friend document, by choosing to use
>>  >>  > ont:UniqueProperty class, licensed inferences
>>  >>  > based on the specification of that class. The conclusion
>>  >>  > that "35" is an :age of :margaret is supported
>>  >>  > by the DAML+OIL spec.
>>  >>
>>  >>  Yes, but not by the RDF spec, and any agent has no business labelling
>>  >>  anything as RDF inferences that are not sanctioned by the RDF 
>>model theory.
>>  >
>>  >Hmm... that's one way to think of 'RDF inferences'.
>>  >It's not one that appeals to me.
>>
>>  Well, sorry to speak plainly here, but tough shit.
>
>I understand you to disagree.

I wasn't disagreeing with your statement about appeal, but wishing to 
convey that phrases like "AAAA inferences" , where AAAA refers to a 
formal language with a defined model theory, are pretty universally 
understood to mean "inferences that are valid according to the AAAA 
semantics". It is pretty hard for me to think what else it could 
possibly mean, in fact.

>  > That is what
>>  'inference' *means* when qualified by the name of a formalism.
>
>It seems to me there are many formalisms that fit in the
>Resource Description Framework, if you mean 'formalism'
>in the technical sense,

Yes, I mean it in a reasonably technical sense. And again, we are 
obviously caught in a severe breakdown of communication, because I 
have always understood the term "RDF" to refer to a particular 
formalism, the one described, oddly enough, in the RDF specs being 
slowly produced by the RDF Core WG. I don't think that interpretation 
is an unreasonable or capricious one, either, unlike yours.

(Strictly speaking, each RDF vocabulary defines a single formal 
system in the Davis sense; but this is just a terminological style 
difference, since the same model theory and notions of inference 
apply to them all. This usage is usual in mathematical logic but not 
in computer science, so I am using the term more in the CS sense, 
where for example we would call Java a language, rather than 
describing each Java vocabulary as a separate formal system. In the 
Davis-style usage for example it would be incorrect to refer to 
first-order logic as a formal system; each first-order vocabulary 
defines a different formal system in this sense. The same semantics 
and rules of inference apply to them all, however. The RDF MT 
document makes a kind of passing nod to the math-logical usage by 
defining an interpretation relative to a vocabulary.)

>ala
>
>% Formal Systems - Definitions
>% (from Ruth E. Davis, Truth, Deduction, and Computation.
>% New York: Computer Science press, 1989.)
>%
>http://www-rci.rutgers.edu/~cfs/305_html/Deduction/FormalSystemDefs.html
>% (c) Charles F. Schmidt
>% Last Modified: Saturday, May 08, 1999 9:07:08 PM GMT
>http://www.w3.org/XML/9711theory/FormalSystem.lsl
>
>But OK, I can see that it's misleading to use 'RDF inferences'
>that way; Peter introduced it into the conversation, not I:
>
>>  >>  Yes, but not by the RDF spec, and any agent has no business labelling
>>  >>  anything as RDF inferences that are not sanctioned by the RDF 
>>model theory.
>
>The agents in question don't label their conclusions as
>'RDF inferences', so I don't see what his point is.

The debate, as I understand it, has been about whether these thingies 
are appropriately called 'RDF agents'. I took this terminology to 
imply that I would be entitled to assume that any inferences they 
draw are RDF-valid. If not, I fail to see what utility there is in 
calling them "RDF agents".

>  > If you
>>  misuse English, then the appropriate thing to do is to ignore what
>>  you say.
>
>I don't think you'll find terms like 'RDF inference'
>or even 'inference according to a formal system' in
>an English dictionary.

You will find the latter, and usages like the former, in an English 
mathematical dictionary.

>  > >  > > I can think of two agents (cwm and Euler) that
>>  >>  > do a lot more than simple entailment, when
>>  >>  > asked to. I think of them as RDF agents.
>>  >>
>>  >>  They are not.
>>  >
>>  >Er... I accept that as your opinion.
>>  >I disagree.
>>
>>  The please tell us what you do mean by the phrase. Right now I have
>>  no idea what you are talking about, and your usage seems to be in a
>>  world of its own.
>
>I didn't introduce the term 'RDF agent' and I feel no obligation
>to define it. When Peter introduced it, I inferred that an
>RDF agent was licensed to conclude simple entailments plus
>other stuff that it's instructed to conclude (e.g. by
>feeding it more axioms on the command line).

Axioms on the command line?? (I have no idea what you are talking 
about here, sorry. Axioms in what language?)

>My position on the matter at hand, media types for RDF/DAML+OIL/OWL,
>is that we should have a media type, application/rdf+xml; documents
>labelled with this media type necessarily license all simple
>entialments that follow from them; but that's not the only
>way that they constrain interpretations; they also constrain
>interpretations according to the specifications of the terms
>used as predicates in them.

Specifications in what language? If the specifications can be in ANY 
language then RDF has magically acquired the ability to express all 
of human thought, literature and science. Which seems to Peter and me 
to be ridiculous, but which you seem to be implying. If we are 
misunderstanding you, please correct our misapprehension.

Pat

-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Thursday, 30 May 2002 14:41:45 UTC