Message-ID: <323724D6.797C@pobox.com>
Date: Wed, 11 Sep 1996 22:45:10 +0200
From: Eric Vereerstraeten <eric.vereerstraeten@pobox.com>
Organization: MYSELF
X-Mailer: Mozilla 2.02 (Win95; I)
MIME-Version: 1.0
Newsgroups: comp.ai
To: jgarzik@pobox.com
Subject: Re: Some basic questions about learning AI
References: <50ikki$o6@havoc.gtf.gatech.edu>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 194.78.41.202
Lines: 44
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!newsflash.concordia.ca!news.mcgill.ca!mcrcim.mcgill.edu!bloom-beacon.mit.edu!news-ext.crl.dec.com!decwrl!enews.sgi.com!news.mathworks.com!newsfeed.internetmci.com!in1.uu.net!nntp.inet.fi!news.bru.tfi.be!news.skynet.be!

Sven Sullivan wrote:
> ...
> Dumb questions:
>
> 1) What type of AI should I lean towards?  There is a plethora of
> available AI algorithms, and some are better suited to certain
> tasks than others.  The AI needs to learn, scan English text (I
> guess either via an NL parse or just scanning for keywords and key
> phrases), and it needs to evaluate the worth (score) of an article.
>
> 2) Where should I start?  Are there any specific source-available
> systems I should look at?  I prefer C++, but I know C, Lisp, and most
> of the other common languages.  There are dozens of books listed in
> the various FAQs.  Are there one or two definitive "AI intro" books?
> Are there a couple related to my news scanning problem that I should
> definitely pick up?
> ...
> Any answers or pointers in this direction will be greatly appreciated.


Dear Sven,

For a program that deals with natural language parsing and with codes available in C on
the Internet, I suggest you to look at the ThoughtTreasure Project at
<http://www.panix.com/~erik/tt/html/tt.html> for the explanations and at
<http://www.panix.com/~erik/tt/html/Code.html> for the source code.

It is of the size of one person working for two years. It's a personal project like
yours. So, don't be discouraged!!!

I've read already some of the main C files myself. It's just fascinating.

If you want some ideas, an unofficial approach to AI, I suggest you to read my web site
at <http://www.wp.com/globint>. Concerning the "score" of an article, here's an excerpt
of my chapter called "Frame":

>>>I've imagined to put a weight factor on words. A piece of text would also have a weight
given by the sum of the weights of all the words.

The text with the biggest sum would have the greatest priority to be memorized. The
weight of each word would depend on the interests of the I-Agent.

For example: when I see the word "Intelligence" in a text, I try to read the rest and to
memorize it.

If I see "Football", I usually don't read the rest (I don't like football). There the
word "Football" has a negative weight for me.
>>>

Happy readings,

---
Eric Vereerstraeten
eric.vereerstraeten@pobox.com

