SUO: Re: Industry takeover
Adam,
I agree that Robert and the other IFFers
have a responsibility to demonstrate the
utility of their approach.
But I can't accept the following point as an
excuse for ignoring the problem of mapping
ontologies:
> I agree, but that doesn't mean that's what
> this group is supposed to be focusing on.
> Often when standards processes progress,
> new needs are discovered, and new groups spin off.
The fundamental core of the SUO charter is to
develop upper-level categories that can be mapped
to whatever lower-level categories are needed
for any specific application. Applied Semantics
has gathered millions of such categories from
many different fields, and all of them are
essential in their domain of applications.
That means that mapping is central to how an
SUO is going to be used. Unless we address
that issue, we are not meeting the requirements.
We cannot dump a bunch of categories on somebody
and tell them to go do whatever they want with
them -- there has to be a methodology for
designing and using suitable tools.
> A number of different DARPA contractors are working
> in ontology mapping. Many are doing so with informal
> approaches, others are using logic, a few are using
> category theory. I'm partial to the logic approach.
> No one seems to have much empirical evidence which
> approach works best.
The simple answer is that you need all of them,
and they must work together:
1. Informal methods of gathering data, editing
the results, and "eyeballing" the categories
to see how they could be aligned are useful.
Nobody is questioning that point.
2. But you also need logic-based methods in order
to guarantee that the results are consistent.
Nobody who understands that point would question
it.
3. Category theory is nothing more nor less than
a systematic set of guidelines for defining
maps from one theory to another and ensuring
that the logical constraints are met. The
reason why the Kestrel people don't emphasize
category theory is the same reason why NASA
doesn't emphasize trigonometry: it's the
foundation for all their calculations, but
it isn't visible in the products.
Compromise: I will admit your point that the
utility of the IF framework must be demonstrated
before it is frozen into an IEEE standard. And I
will also admit that the best way to demonstrate
its utility is to apply it to real data and see
how or if it proves to be useful.
Therefore, I recommend that we ask all three groups,
IFF, SUMO, and OpenCyc, to cooperate in determining:
1. How much of IFF is useful and should become
part of whatever standard the SUO develops.
2. How much of SUMO is useful and should become
part of whatever standard the SUO develops.
3. How much of OpenCyc is useful and should become
part of whatever standard the SUO develops.
Right now, we have three groups, each of which
is convinced of the importance of its own project,
but has doubts about the value of the other two.
If all three groups continue to work independently,
none of them will ever convince the others (or anyone
else, for that matter) of the value of their own
contribution.
The point of my motion is to ask all three groups to
admit the possibility that the others have something
useful to contribute and to collaborate with them
to see they can work toward some useful combination.
If all three groups make an honest effort to work
together and they find that one or two of the others
are ineffective, then the final standard will not
include much, if any, of the weaker one(s). All I'm
suggesting is that they begin the effort with an
open mind and let the results speak for themselves.
John