Alex,

There is a big leap from rule-based NLP to statistical, machine-learning
based NLP.  Your group has taken that leap, which is great.  Incorporating
feedback from human chatters is a nifty variation on the stats/ML approach,
but still in my view within that approach.

I think there's another at least equally big leap to be made from the
statistical, machine learning based NLP to NLP that is based on grounding
semantics in nonlinguistic reality.  This step opens a big AGI can of worms,
but is very necessary.  And I'll be very curious if you guys choose to take
this leap at some point...

> Using short sentences gives good results, now its time to make the
> sentences longer. More world-knowledge is needed.

This is exactly the point.  I don't think incorporation of world-knowledge
will ever work very successfully if one is using ONLY ungrounded,
expert-system style knowledge.  I think some knowledge gained via experience
in a nonlinguistic world is needed.  And the integration of this experience
w/ linguistic capability requires a whole different sort of architecture
from a stats/ML based or rule-based NLP or chat system.

Nevertheless, it sounds like your work is going in a good direction and it's
great that you've found a way to do some interesting science slowly moving
in the direction of AGI, while making money too...

-- Ben G

-------
To unsubscribe, change your address, or temporarily deactivate your subscription, 
please go to http://v2.listbox.com/member/

Reply via email to