Re: Structuring relations for the R.O.

John F. Sowa (
Mon, 30 Sep 1996 19:24:02 +0500

Kavi Mahesh writes:

- are we going to use "subrelation" to taxonomize qualities (aka
attributes) or do we need a separate "subquality"? E.g., how do
we connect length and height to size?

This opens up a very large issue, namely the question of first-order
concept types like red, large, and heavy compared to second-order
concept types like color, size, and weight. It also gets into
definitional issues of volume or area in terms of other dimensions.

For information retrieval, a vague association between length and size
may be useful. But for precise reasoning, this whole topic must be
formulated very carefully.

- are taxonomic links going to be bidirectional? For example, in the
case of instance-of, are we going to list all the instances in
the parent? Or, is this something that is to be derived by the
API/other programs?

There is an implementational issue as well as a representational issue.
In logic, every possible relationship is implicitly available to the
inference engine, so any such question could in principle be answered.
There is no need to clutter up the notation with inverses and such.

In an implementation, there may be serious computational implications.
For example, asking for all instances of planets in the solar system
is no big deal. But asking for all instances of bacteria on earth
may be hard to answer -- even if you had an up-to-date database that
listed them, it would be obsolete by the time you printed it out.

John Sowa