Re: Representing quantification in RDF

From: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>

> If all you want is syntax, with active components being black boxes, then
> there is nothing to prevent you from using RDF, or S-expressions, or even
> C++ syntax to encode quantification.

Fine.  I would, however, like to point out that the level of detail at which
"active components"  become black boxes is quite arbitrary and is up to the
implementer.  I implemented a system once where "active components" were
black boxed at the machine instruction level.

> If you want to ``specify'' or
> ``represent'' or ``describe'', then you need much more.  In particular,
you
> need some way of saying what the syntax means.  RDF does not provide a
> mechanism for this, nor do diagrams.

Well perhaps we have no agreement as to what "specify", "describe",
"represent" and "mean" means.  Especially since my version of these concepts
all require an indirect object to be at all meaningful.  For example:  {"X"
represents Y } is fairly meaningless to me; but {"X" represents X; to (some
active process)} starts to describe the predicament.  Please refer to the
diagram [1] where I have superimposed the RDF syntax on Sowa's meaning
triadic described in [2].  The nodes in red may be implemented inside the
system with RDF syntax or (in the case of the cat) will be impossible to be
implemented inside the system.

[1] http://robustai.net/mentography/semiosis.jpg
[2] http://www.bestweb.net/~sowa/peirce/ontometa.htm

Seth Russell

Received on Sunday, 6 May 2001 19:04:50 UTC