Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!eecs-usenet-02.mit.edu!news.kei.com!nntprelay.mathworks.com!howland.erols.net!usc!news.isi.edu!gremlin!shomase!jbarnett
From: jbarnett@nrtc.northrop.com (Jeff Barnett)
Subject: Re: Tree of Expert Systems?
Message-ID: <ECnr64.9Kv@gremlin.nrtc.northrop.com>
Sender: news@gremlin.nrtc.northrop.com (Usenet News Manager)
Reply-To: jbarnett@charming.nrtc.northrop.com
Organization: Northrop Automation Sciences Laboratory
References:  <33B942D5.51C9@esumail.emporia.edu>
Date: Tue, 1 Jul 1997 20:49:16 GMT
Lines: 20

In article <33B942D5.51C9@esumail.emporia.edu>, R Jones <jonesrob@esumail.emporia.edu> writes:
|> Has anyone assembled a collection of expert systems into a
|> tree?  Each expert system would, itself, be a leaf and the 
|> "tree" would direct the user to that individual expert appropriate to the
|> question/subject at hand. Does this seem like a useful idea?

There is, to me, an interesting reflection based on this idea:
Simple trees have proved to be an INADEQUATE structure to use
as a knowledge base when the domain is complex.  Much more
exotic/convoluted/deeper structures always seem to be necessary.

So the answer to the posters question, in my mind, is based on
how complex the domain of expert-system descriptors is.  My
guess is that the idea will not work very well once the domain
of competence reaches a few hundred and there are many examples
that play ball in the same field.  I don[t think it's good
enough to identify MYCIN from the keyword "medicine" -- there
are too many medical examples and "everybody" knows that anyway.

Jeff Barnett
