[topicmapmail] Superclass-subclass indentation in the Omnigator

[Robert Barta]

On Sun, Dec 29, 2002 at 08:02:17PM +1300, Conal Tuohy wrote:
> Tom Passin wrote:
>   Perhaps we should ask whether a higher-order type can have individual
> concrete instances ("Pookie is a mammal") or whether that construction,
> although common in natural languages, needs to be banished from our
> ontologies.
> Why should we banish this construction from our ontologies?

That would be theoretically nice, but is simply not practical, IMHO.

It would mean that some GOD (Grand Ober Directory) would decide which
topics are "instance-able" and which are not. Difficult, even in a
world of monopolies.

> If we know that Pookie is-an-instance-of Gorilla, and also that Gorilla
> is-a-subclass-of Mammal, then we can simple DEDUCE that Pookie
> is-an-instance-of Mammal. Can't we?

I see actually no problem in someone saying "Pookie is a mammal" as
long as this is consistent with any map I merge:

map A:
-----------------------------
pookie is-a mammal

map B:
-----------------------------
pookie is-a gorilla

gorilla subclasses ape

ape subclasses mammal

would merge into

A + B
-----------------------------
pookie is-a mammal is-a gorilla

If "reasoning" is switched on this still can be simplified to

pookie is-a gorilla

\rho