[Somewhat Off Topic] AI Contest

François Pinard pinard at iro.umontreal.ca
Sun Aug 5 20:48:36 CEST 2001


[Roman Suzi]

> To be good for AI, language must have logic programming support.

Hello, Roman.

"logic programming"?  Do you merely mean using some logical inference
engine, or more than that?  Or would a regexp matcher fit the bill?

> (I wonder if Guido thinks in this direction 'cause this requires adding
> new keywords ;-)

You say "requires"?  I'm curious about why new keywords would be needed.
Python would not be expressive enough as it stands?

> the solver engine comparable to Prolog is to be written.

Nobody did this yet?  I thought a lot of little logical inference engines
existed here and there, I did not paid much attention to these, recently.
Traditionally, LISP is considered good for AI, without having built-in
goal seekers.  I would not fear using Python for AI projects...

I worked in AI fields for some while, before doing Operation Research, later.
I was rather surprised to find out that AI and OR were sometimes addressing
similar problems deep down, yet formulated quite differently.  The OR
side was much more humble, and often reached better and faster results.
With a broad enough perspective, FORTRAN is a good AI language! :-)

P.S. - The second Prolog interpreter was written in FORTRAN.  Maybe the
first was written in Algol-60.  Or, maybe, am I thinking about Systèmes-Q?
This is all old.  In any case, both were designed by the same guy: and if
the later Prolog was much speedier than Systèmes-Q, it lacked their elegance.

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard




More information about the Python-list mailing list