Re: CG: Re: Top level ontology

Sergei Nirenburg (
Tue, 02 Dec 1997 15:12:12 -0700

"JFDelannoy" writes:

> > And in response to Sergei, I don't see any clear distinction between
> > commonsense categories as expressed in natural languages and the highly
> > formalized categories of science, especially mathematics. All of the
> > scientific terms (including number, space, time, etc.) are refinements
> > and extensions of the commonsense terms. And the commonsense terms
> > tend to be much more stable over thousands of years, while the scientific
> > terms tend to shift their meanings with every passing fad.
> Scientific terminology uses common terms, but metaphorically:
> - for particles: spin, charm, colour,
> - for numerical analysis: convolution, matrix inversion, fixed point
> - in astrophysics: black hole
> - in linguistics: tree, trace, stem
> The notion of stability cuts both ways: the commonsense vocabulary is more
> stable across century (diachronically), but at all point it is often
> polysemous, or has interindividual and intergroup fuzziness (how open
> should a door be to be called ajar? How big is a big book ? How many
> grains of sand make a sandpile? and other sorites).

Yes, many of the scientific terms can be considered to be additional senses of
"commonsense" words. However, it is, in my opinion, dangerous to read too much
into any similarities. You know, "salt" in chemistry subsumes "soap" though
this is a violation of a commonsense view, right? And "salts" in chemistry
have no privileged sense connected with either smelling or baths.

I have always defended the position that there is no one-to-one or even
near-one-to-one correspondence between NL lexical units and ontological
concepts in an ontology designed to help computers process natural language

Specifically because of polysemy as an inherent feature of natural language. I
hypothesize that it is inherent because in interpersonal communication the
commodity which is at a premium is time (the ability to communicate *fast*),
since the disambiguating devices (people) are very very good. Computers are, to
put it mildly, inadequate disambiguation devices, so they need much
finer grain-sized description units, each of which is, with luck, not ambiguous
at all.