RE: What do the ontologists want

>pat hayes wrote,
> > As far as I understand what is meant by 'reification' in this
> > context, I see only a very limited utility, basically things like
> > tagging a string/expression with information about its source,
> > time-stamping and so on.
>
>I think this is a little hasty. Quotation and 'reification', in your
>scare quote sense of abstract syntax rather than propositional
>content, have some very important roles to play. To borrow some
>examples from Tarski and Davidson,
>
>  "Snow is white" is-true-in-english iff snow is white
>  "Schnee ist weiss" is-true-in-german iff snow is white
>
>or better,
>
>  "Snow is white" means-in-english that snow is white
>  "Schnee ist weiss" means-in-german that snow is white
>
>The first example in each of these pairs baffles first year philosophy
>undergraduates, but the second makes things somewhat clearer: where
>the quoted expressions are drawn from an given object language the
>respective truth and meaning predicates can be used to set up
>correspondances target (I won't say meta- here) language. Fill this
>out sufficiently and toss in some syntactic structure on the quoted
>side

That is quite a big toss, however.

>and you have a translation scheme for the object language to the
>target language.
>
>Bearing in mind that a lot of the stuff that people will be wanting
>to do with RDF is set up mappings between local and non-local
>vocabularies this strikes me as being of more that 'very limited
>utility'.

Fair enough. This is the very kind of reason why we wanted to keep 
full meta-descriptive abilities in KIF. However, the pragmatists all 
argued vehemently that anyone who seriously wanted to specify 
translations or syntactic mappings would use LISP or Python or some 
such and would be crazy to want to put such stuff in the actual 
logical language itself, and I do see their point.

(For the record, the kind of usage that really seems to be a win for 
reification in logical KRep is allowing the langauge to extend its 
own syntax, so that for example one could write axiom schemas which 
describe syntactical restrictions on certain classes of identifiers 
and relate them to other expression forms. Then for example a 
reasoner could conclude that since an ontology was written by some 
source, any identifer in it starting with an up-arrow symbol must 
denote a transitive binary relation, say.  But this seems to me to be 
pretty exotic stuff compared to the big-shallow-DBs-and-catalogs kind 
of applications that seem to be the main driving force behind DAML 
and RDF, and the reasons why RDF syntax needs, apparently to be kept 
so 'simple'.)

>But don't take my word for it: cp. Tarski, "The concept of truth in
>formal languages".

My bible, my dear fellow. But then Tarski would probably have had 
kittens if he had to deal with XML.

Pat Hayes

---------------------------------------------------------------------
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 Tuesday, 15 May 2001 19:07:26 UTC