[Top] [All Lists]

Re: [ontolog-forum] Toward Human-Level AI

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Tue, 03 Jun 2014 23:09:10 -0400
Message-id: <538E8DD6.2060805@xxxxxxxxxxx>
One of the major advantages of decision trees is that every such
tree can be translated automatically to a huge nest of if-then-else
statements.  Then that nest can be compiled into a very efficient
program in any programming language that supports if-then-else.    (01)

Paul Compton, who was a colleague of Quinlan's, coined the term
Ripple Down Rules (RDR) and developed tools and methodologies
for using them:  http://www.cse.unsw.edu.au/~compton/    (02)

The RDR tools have been popular for medical informatics.  Some of
those applications have created decision trees with thousands of
branches.    (03)

Some people have developed controlled English interfaces to those
tools.  Users enter English-like phrases and simple arithmetic
expressions for the conditions and conclusions.    (04)

Whenever the system makes a mistake, users type the correct
conclusion it should have drawn.  Then the system revises the
RDR rules.    (05)

Nobody would ever want to read the entire tree with thousands
of nodes.  But the path to any conclusion is usually rather short
-- about 4 to 6 steps.  Since all the conditions and conclusions
are in controlled English (with some arithmetic), the system can
explain any decision by printing out the steps it took.    (06)

John    (07)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (08)

<Prev in Thread] Current Thread [Next in Thread>