Re: [MLUG - DISCUSSION] Re: [MLUG] AI - any suggestions?

[Kmicic <EMAIL:PROTECTED>]

Michael wrote:
 
> > seems to me you're concentrating on things of little relevance to nlp
> > here. mapping tokens into dictionary words is a largely mechanical task.
> 
> Umm and what would you do. Map them to nothing whatsoever? A word without
> meaning means nothing to nobody. You can work off pure semantics but then
> you'll just be babbling like ELISA.

please, read what i said again. i said tokenizing is a mechanical task
of little relevance to actual parsing. i did not say it was not
necessary to tokenize before you begin parsing, simply that you were
devoting a lot of space to a menial portion of the nlp task while
ignoring the important issues.


> > accept your claim that "parsing nl is easy". if i am correct, you're
> > suggesting a phrase lookup kind of nlp system here (perhaps with some
> > lematization, ie stripping words into their word stems). this is all

> 
> All your doing is mapping from a very generic representation of a word
> down to an easy to work with form. Known words will already have a key in
> memory and likely have some sort of definition as to if they are a verb,
> noun, etc.. regardless as to if your using a provided db or one generated
> by a learning process (or both). Mapping the words to keys in this manner
> saves effort in processing and lets you try to match a known grammar.

if you map the word "left" in the example i gave to "a very generic
representation," it'll be an error because in one part it is a verb
(past participle of leave), and in another a noun (side opposite of
right). hence your lookup-based system will need to have 2 separate
entries for the 2 contexts (and many other contexts) if you don't do any
grammatical parsing to figure out their place in the sentence. hence
your system will grow very large very fast.
 
> What evidence do you have that it breaks down in a large system? I've had
> web agents that could take very generic requests and go find the needed
> files for the user etc. Also had bots with much higher understanding
> abilities than anything else I've seen.

why not present those mechanical geniuses to the public instead of
hiding them in your closet? otherwise your claim is unverifiable.


> > - given the fact that I have a truck, 4x4, a bicycle and a horse, how
> > many cars do i have?
> > or
> > - i left the passenger-side window down today, and it was raining. is
> > the left car seat wet or dry?
> > this sort of tasks with the system you described.
> 
> It depends on the processing engine you put behind the parser. With a
> simple scripted bot then unless you've designed it to answer such
> questions then yes it'll fail. If these were the questions you were
> targeting then it shouldn't be hard to work with. 

please explain how.

 
> I'm not explaining processing at all. The discussion was on parsing. I

i suggest you look up definition of parsing. it has to do with
language's grammar and putting structure on a stream of tokens. getting
the stream of tokens (which you devote so much time to discussing)
technically is not even a part of parsing. i am not much for holding
someone to strict definitions in a discussion, so i gave you benefit of
the doubt and thought you meant "nlp is easy"

> processing engines. Since the original discussion was on IRC bots I'll
> assume we can simplify it down to a scripted bot..

i was under impression the original discussion was NOT on irc bots. you
have stated that "parsing nl is easy" and then proceeded to give irc
bots as an example.

 
[snip pseudocode] 
> Does that make any sense. Basiclly your usually looking for certain words
> in the stream.. sometimes in order.. sometimes out... and will usually use
> a reg expr to check.. then you want to check the first grammar key in the
> stream 

what is the first grammar key?

so that you can know what type of sentence you were working
> with. This would mean checking if it's a statement or a question, if a
> word was being used asa  noun or verb, etc. Then working on both those
> conditions you'll probably want to output your list of words so that your
> function can do something with them..

also not sure what you mean here. what is "your list of words"? perhaps
you need a little diagram of the input flow to clarify your description. 

 
> If you have studied neural nets then the concept of feeding them strings
> of keys (symbols) shouldn't seem very surprising.

i have studied neural nets. frankly i don't know of an obvious way to
employ nn in nlp. perhaps you can explain to me by specifying what kind
of input you'd feed to nn, what kind of output you'd expect on each
output node, and how you'd define error for each output node.

/paul
--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/