RE: What do the ontologists want

>At the recently conducted conference in Hong Kong, I was puzzled by 
>the fact that in all the official SW sessions RDF was being pushed 
>almost as the completed essence of the Semantic Web itself, whilst 
>the many difficulties in RDF's capacities -- which numerous of the 
>participants were surely aware of through these threads and other 
>forums -- were never really aired.
>
>In informal corridor discussions with some of the individuals 
>heavily involved in the ontology community, it was explained to me 
>that they regard RDF simply as a tool for expressing graph relations 
>-- i.e. the triples encode nodes and the arcs between them -- 
>without any of the other promised capabilities, whether they be 
>semantic, model-theoretic, reification, and so forth. No less and no 
>more.
>
>If this is indeed the only "real" use the ontologists are intending 
>for RDF, perhaps the question of whether RDF is an efficient 
>mechanism for that purpose as well might deserve some discussion.
>
>
>Cheers,
>
>Ziv
>
>
>PS My apologies if this is leading too far from the original thread


Not at all: I think you have put your finger right on the button. 
There is a mismatch between the reality and the rhetoric of RDF.  As 
a general-purpose graph-structure-encoding formalism, it is just 
fine. (It has a truly awful surface syntax, but everyone agrees that 
is a crock and needs replacing.) But it hasnt been sold as that: it 
has been sold as a universal knowledge/information representation 
language, with a clear semantics which is both utterly simple 
(relational triples) and simultaneously universal, post-Goedelian, 
trans-Tarskian and magically universal, due to the Power of 
Reification. That is why it is going to be the, I don't know, the 
magical essence of the Semantic Web, and why W3C seems to be so 
committed to it. And this is just plain bullshit, as anyone who knows 
almost anything about Krep will immediately tell you. (Most people 
who know anything about Krep have either refused to get involved with 
RDF or have given up and taken a shower.)

What we need to do is to first disentangle the reality from the hype, 
and then use some professionalism in trying to actually do some of 
the Krep work that the hype claims has already been done, but do it 
properly.  DAML has been trying to make a start in that direction, 
for example. But it isnt going to be easy, and there aren't any 
magical solutions. Krep is a lot harder than hypertext markup.

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 Wednesday, 16 May 2001 22:08:28 UTC