Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: notesfiles
Path: utzoo!watmath!clyde!cbosgd!ihnp4!qantel!hplabs!hp-pcd!orstcs!tgd
From: tgd@orstcs.UUCP (tgd)
Newsgroups: net.ai
Subject: Re: inductive learning in expert systems
Message-ID: <14100004@orstcs.UUCP>
Date: Sat, 3-Aug-85 15:53:00 EDT
Article-I.D.: orstcs.14100004
Posted: Sat Aug  3 15:53:00 1985
Date-Received: Tue, 6-Aug-85 11:37:53 EDT
References: <770@utcs.UUCP>
Organization: Oregon State University - Corvallis, OR
Lines: 36
Nf-ID: #R:utcs:-77000:orstcs:14100004:000:1501
Nf-From: orstcs!tgd    Aug  3 11:53:00 1985


Yes there are a number of applications of inductive inference
techniques to the development of expert systems.  The first, and
perhaps best, was Meta-DENDRAL, which developed rules for the
DENDRAL expert system.  Also, the AQ11 system has been employed
to develop rules for a plant pathology expert system.

A few companies are marketing inductive inference programs
for this purpose.  Radian corporation is marketing something
called RULEMASTER, which forms rules from examples.  Human Edge
is marketing Expert Ease, while builds decision trees from
examples.  Both of these systems, I believe, employ a method
developed by Ross Quinlan called ID3.

References:

Meta-Dendral:  See the AI Handbook, Volume 3.  Also,  Buchanan & Mitchell,
in Waterman and hayes-Roth, Pattern-Directed Inference Systems, Academic
Press.

AQ11: Michalski, R. S., and Chilausky, R. L., Learning by being told
and learning from exapmles: an experimental comparison of two methods of
knowledge acquisition in the context of developing an
expert system for soybean disease diagnosis.  International Journal of Policy
Analysis and Information Systems, Vol. 4:125-161, 1980.

ID3: Quinlan, J. R., Learning efficient classification procedures
and their application to chess endgames, in michalski, Carbonnell, and
Mitchell, Machine Learning, Palo Alto, Tioga Press, 1983.

See the latest Ai magazine for adds from Radian and Human Edge.

--Tom Dietterich
  Oregon State University
  orstcs!tgd
  tgd%oregon-state@csnet-relay