----- The following addresses had delivery problems -----
<olsen@piano.stanford.edu> (unrecoverable error)
----- Transcript of session follows -----
554 MX list for piano.stanford.edu. points back to cdr.stanford.edu
554 <olsen@piano.stanford.edu>... Local configuration error
----- Original message follows -----
Return-Path: agents@aic.lockheed.com
Received: from soleil.lmsc.lockheed.com (soleil.lmsc.lockheed.com [129.197.2.1]) by cdr.stanford.edu (8.6.8/8.6.6) with ESMTP id JAA13890 for <olsen@piano.stanford.edu>; Mon, 3 Oct 1994 09:57:47 -0700
Received: from aic.lockheed.com (goliath.rdd.lmsc.lockheed.com [129.197.131.53]) by soleil.lmsc.lockheed.com (8.6.9/8.6.9) with SMTP id JAA03506 for <olsen@piano.stanford.edu>; Mon, 3 Oct 1994 09:57:43 -0700
Received: from freedom.aic.lockheed.com (freedom.rdd.lmsc.lockheed.com) by aic.lockheed.com (4.1/SMI-4.1/AIC-PostOffice-Brent-930416-01)
id AA13449; Mon, 3 Oct 94 09:56:25 PDT
Errors-To: rjohnson@aic.lockheed.com
Received: from (localhost) by freedom.aic.lockheed.com (4.1/SMI-4.1/AIC-Client-Brent-930416-01)
id AA00629; Mon, 3 Oct 94 09:56:22 PDT
Date: Mon, 3 Oct 94 09:56:22 PDT
Errors-To: rjohnson@aic.lockheed.com
Message-Id: <"swan.cl.cam.:206170:941003160730"@cl.cam.ac.uk>
Errors-To: rjohnson@aic.lockheed.com
Reply-To: agents@aic.lockheed.com
Originator: agents@aic.lockheed.com
Sender: agents@aic.lockheed.com
Precedence: none
From: Ralph Becket <Ralph.Becket@cl.cam.ac.uk>
To: olsen@piano.stanford.edu
Subject: Re: What's the Best Intelligent Agent Software?
X-Listprocessor-Version: 6.0b -- ListProcessor by Anastasios Kotsikonas
X-Comment: Mailing List for Software Agents
That's a tricky one - the agents field is still quite immature.
I think it really depends on what you want to do. I suggest you
have a look at CLIPS (available as freeware, it's a superb rule
based shell that compiles down to C) and DynaCLIPS (not freeware,
but I don't think it costs a bomb). DynaCLIPS is a set of
blackboard extensions to CLIPS allowing you to build DAI type
systems. Part of the problem is that:
i) there's not much `general' intelligence around, since
intelligence is largely knowledge driven, so it's hard
to come up with impressive domain independant agents; and
ii)it's hard to build even general purpose agents without
much intelligence, since different applications require
different representation languages.
Best thing to do would be have a look at (Dyna)CLIPS and post to
comp.ai asking other people's opinion - I'm afraid I'm not an
expert on what's available!
Best of luck,
Ralph