RE: SUO: ontology as science
John F. Sowa wrote:
[snip]
> Cyc supports contextual differences with microtheories, and
> we need a similar modular mechanism in any proposed SUO.
>
> John Sowa
I've read the Cyc stuff about microtheories, but it isn't
exactly a piece of tutorial elegance. IOW I don't understand
the motivation behind it, the broad points of implementation,
or the alternative choices that could have been made instead.
Certainly context is specific to a conversation, or a problem
to be solved, or a lesson to be taught. But isn't context
also decomposed into other (sub)contexts pretty much like
OOD handles scopes? The present OOD methodologies don't deal
well with scopes, using the dot notation to reach any object
from any other object. This isn't a very good solution
because many separately designed units have to have access
to nearly all the other units in a large program. Polymorphism
is a start at improvement, but only just barely.
It would be nice if there were a more formal way to represent
contexts, nestings, decompositions, and so forth.
In XML, the namespaces are no better than in OOD. I know of
no existing solution to this problem, but it lies at the heart
of the exponential nature of software testing and debugging.
A change in one unit can cause zillions of changes in other
units because references from the other units are not very
transparent to the design of the changing unit.
Does anyone have a better solution they're willing to share?
Thanks,
Rich