SUO: RE: Enhancing Data Interoperability with Ontologies...
John F. Sowa wrote
> Rich,
>
> I was trying to make the point that
> neither natural languages nor artificial
> languages guarantee 100% successful
> communication.
I understand that John, and its what motivated
me to ask John Bateman for the comparison if
he has it available. Out here in the business
world, I can't expect perfect responses
from people, and I'm willing to approximate
the results I need where perfection isn't
feasible.
> RC> So a wide comprehension NL question answering
> > system isn't on the horizon, and would take
> > an enormous investment to build up a knowledge
> > base of world knowledge.
>
> That's not what I said. What I was trying
> to say is that perfect comprehension is
> unattainable. And even if you use a perfectly
> unambiguous source language, there is no
> guarantee that what the author said is
> what the author meant.
Well, perfect comprehension will probably always
be unobtainable. And what goes on in the author's
mind will still be something of a guess for a
program as well as for you and I.
> RC> Is that approach feasible?
>
> A lot of things are feasible. Some things
> work fairly well now, but nothing is as
> good as we would like to see.
>
> Summary: There's a lot of research left
> to be done, a lot of trial and error is
> necessary, don't expect any magic solutions
> to everything, but we may get partial
> solutions to some things that are somewhat
> better than what we have.
That's what I mean. Do you have any idea
(ballpark is fine) how well a vertical
domain question answering system would do
in getting domain knowledge interactively
from a domain expert using DRSs, WordNet,
and an ontological structure based on
WordNet concepts to begin with?
As it gets domain knowledge, would it be
able to make reasonably accurate models
that extend the primary ontology until
it covers that vertical domain?
This was a goal back in the eighties that
was considered impossible then by most
people willing to pay for technology
projects. Do you think the goal has become
feasible now that we have more theory,
more computing power, and more practical
linguistics experience?
> John
Thanks,
Rich