> Hi everybody,
>
> I agree basically, that non IT folks should definitively participate
> in that kind of projects. The thing is usually that "we" should
> show them what we are able to do and they take leadership
> then and determine what to do with that.
Heheh - you are assuming that "we" are all "just" IT people!
I did my initial degree in English, which included the reading of much
cannonical modern philosophy, psychology, and metaphysics, to name
but three. I was most fortunate to study my second degree, in AI,
at Sussex, where all (if I'm not mistaken) of the AI staff have
at least MAs, and usually doctorates, in non-IT subjects (as above).
This creates a fantastic atmosphere of mixed ideas, and does seem
to prove that whilst any reasonably intelligent person can pick up
programming in most languages in a year or so, it takes much more
work to find something to program - both an indea to impliment,
and a means of implimentation.
> Why don't we try to implement a chatterbot based on conversational rules,
> and an underlying common knowledge base and try to determine current
> and shifting domains. The project I have seen until now are limited to
> specific domains. (Because that's much easier :))
Exactly, and easy though it may seem, it is still proving to be extremely
difficult. Limiting domain limits linguistic ambiguity, which is a real
boon!
> That would be a big challenge.
And, more importantly, expensive in time....
> I have seen years ago a publication where these rules for "W*" questions
> (what, who, where, why ...) were discussed but I can't remember anymore
> where it was. Does anybody remember?
Not me, sorry, but I am sure there are many many such discussion out there.
Lee
