Re: A modest proposal for reforming RDF

Drew McDermott wrote:

>    > I don't think it needs to look
>    > radically different from RDF, but it does need to give up the graph
>    > model.
>
>    Why?  I can think of several ways to represent logical statements in RDF
>    and all of them need the graph model.  A logical statement is a tree, and a
>    tree is a special case of a labeled directed graph.
>
> This is a very puzzling view.  If this is the way we are supposed to
> think of it, then RDF becomes completely unimportant.  The relation
> between RDF and the language we actually care about is about the same
> as the relation between linked lists and predicate calculus.

Well that's not too far from my view.  But RDF *is* a
standardization; so that our diverse representations have a
level of interpretability that they wouldn't have if we just
said "use linked lists".  RDF gives us a structure on which
we can hang content .. then the URIs gives us content
identity ... the combination gives us interpretable content
.. with no commitment to any particular epistemology ... i
like it :)

> And yet
> the discussion seems to constantly assume that RDF places actual
> constraints on the language somehow.

RDF places a number of important constraints  on the
language .... binary relations being perhaps the most
dramatic ... and that a triple cannot itself bean object is
another troubling one.

> If it doesn't, then what is the
> language exactly?  If the answer is "DAML," then I don't think I've
> seen an actual DAML formula yet.  I've seen DAML *ontology* and I've
> seen RDF *content* (using symbols defined in DAML), but I haven't seen
> *DAML content,* and I don't believe anyone has suggested what it would
> look like.

I didn't think there was any content in a logical formula. 
But I think we get content by using logical formula ... for
me, how we do that hangs off the schema dimension ... off
the labels on the arcs  ... and personally I'd like to see
it look pretty much like OO. I sketched this once ... see
[RDF OO].

Incidentally you never did tell me "why?" under you view we
would need to give up the graph model.  I'm still curious to
grok that view.

[RDF OO]
http://lists.w3.org/Archives/Public/www-rdf-logic/2000Nov/0134.html
Seth Russell

Received on Wednesday, 20 December 2000 13:54:30 UTC