Re: followup from telecon (integration of XML and RDF)

From: Pat Hayes (
Date: 12/03/01

>At 03:46 PM 11/29/2001, Pat Hayes wrote:
>>>thanks for your answer.
>>>I'm still not clear if I could represent an interpretation graph
>>>in RDF, given the right vocabulary.
>>My reply would be: suppose you can. So what? Ie what significance 
>>would you attribute to the ability to describe such a structure in 
>my understanding is, that RDF can serve as a universal data model - thus
>representing any structured data.

Yes, but representing data, and representing one interpretation of 
some data, are not the same thing. I understood you to be talking 
about the latter, not the former.

>Thus investments in RDF can be
>justified by the reusability of the results.
>If this is true, tools suitable to store and query RDF data can be 
>used for XML
>documents as well, if Peter's interpretation of XML documents is applied.

? How? Just representing one interpretation does not represent all 
the content of the document. You would need, in the general case, 
infinitely many such representations (disjoined) to represent the 

>If RDF is not suitable as a universal data model, I would like to figure out
>why not.

Well, come come; you are enough of a logician to work that out for 
yourself, surely. Obviously a simple existential-conjunctive logic 
with binary relations is not sufficient as a universal data model.

>>(PS. BTW, it depends on what you mean by 'describe', by the way. If 
>>you mean, can you make an RDF graph that is isomorphic, then of 
>>course you can if the interpretation is finite, and not if it is 
>Hm, we are talking only about finite amounts of data, right?
>>If you mean, can you write some RDF so that SOME interpretation of 
>>it is isomorphic, the answer is yes, trivially; if you mean, write 
>>some RDF so that ALL interpretations are isomorphic, the answer is 
>>no. If you mean, so that all interpretations in some class of 
>>interpretations are isomorphic, (eg all 'minimal' interpretations 
>>in some sense), then say what class you have in mind.
>My question was if there are computable functions f, fi, with f maps an
>XML interpretation to an RDF graph, such that fi o f = id.
>Is it the last case?

The odd thing to me is to even be in the business of mapping 
interpretations to graphs. Why would one seek to have a computable 
mapping from semantics to syntax?


IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax

This archive was generated by hypermail 2.1.4 : 04/02/02 EST