Re: What do the ontologists want

From: "pat hayes" <phayes@ai.uwf.edu>

> >From: "pat hayes" <phayes@ai.uwf.edu>
> >
> > > But whatever you call it, the point is that the *logical* syntax in
> > > this case is NOT the triples, but these more complicated structures
> > > that are being implemented as sets of triples. So we need to somehow
> > > specify the rules for what counts as being well-structured (not every
> > > set of triples will be) and we will probably need a few
> > > datastructuring primitives (like end-of-chain markers, cf. Lisp NIL.)
> > > In other words, we will have to do a little work. Just a tad, but we
> > > will need to do it. Sorry if that's not anarchic enough. (There isnt
> > > anything odd or exotic about all this, let me emphasize: its is just
> > > ordinary bread-and-butter datastructure design. We really do teach it
> > > to our undergraduates.)
> >
> >Yes triples combine to form more complex structures.  The individual
triples
> >that compose those larger structures cannot be isolated and mean
anything.
> >But it is a slippery slope as to where you draw the line ... and i think
> >that is true of any  kind of logic that deals with reality.
>
> Most (all?) of the logics I know draw this particular line very
> sharply, so I don't see it as very slippery at all.

Obviously you dont see the slope .. oh well!

> >In other words
> >it is the behavior of the entire graph (model) that has meaning, not an
> >individual piece.
>
> Well, scrub round 'behavior' (graphs don't behave. At best, an
> interpreter of a graph might behave, but thats not the meaning.).

Ok point taken:  Graph+interpreter yields behavior.

> And
> the basic idea of 'language' is that the meaning of a complex
> statement can be somehow composed from the meanings of its parts.
> (Thats what people often call the principle of compositionality, and
> it applies to things like diagrams and maps as well as text.)

Well I think there are a number of 20th century semanticist that would
dispute that rather vigourously.  But my point is that if you have a system
where graph+interperter yields behavior, then the easiest way to predict the
behaviro of the system after you remove selected statements is to run the
interperter on the mutated graph.  I don't see where the principal of
compositionality comes into play.  The interrelated linguistic statements in
a system are not like a pile of bricks ... nor will they ever be, regardless
of the language.  Me thinks such a requirement is an absurbity.

> > The simplest case of this that I can think of is the
> >concept 'give' ... {John gave candy to Mary} .... I have coded this in
> >tripels for you amusement [1].  It is just this kind of conceptual
> >dependency (as Shank called it) that triples do well.
>
> I dont see that triples do it particularly *well*. They can do it,
> but so can almost anything that can encode directed graphs.

Well actually triples will outperform fixed place predicates in a
distributed KB like the Semantic Web.  (See Stephen Reed's post today about
CycL) One reason is that we can add slots to concepts rather easily (see
[1]) without breaking the whole structure,  but to add a term to a fixed
place predicate we need to rewrite all the old instances. Doing that is
absolutely not practical in the Wild West of the Web.

[1] http://robustai.net/mentography/conceptualDependency.gif

Seth

Received on Friday, 18 May 2001 19:54:53 UTC