Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.ecn.bgu.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!tank.news.pipex.net!pipex!news.mathworks.com!news.kei.com!simtel!harbinger.cc.monash.edu.au!lugb.latrobe.edu.au!latcs1!johnz
From: johnz@latcs1.lat.oz.au (John Zeleznikow)
Subject: Re: (ab,in,de)duction
Message-ID: <DEIIu1.6s2@latcs1.lat.oz.au>
Organization: Comp Sci, La Trobe Uni, Australia
X-Newsreader: TIN [version 1.2 PL2]
References: <426u19$6hf@midgard.calvacom.fr>
Date: Thu, 7 Sep 1995 01:58:48 GMT
Lines: 51

Maurice Clerc (mcft10@calvanet.calvacom.fr) wrote:
: I read recently an interesting paper in Ai Communications "A 
: Framework for Abductive Rule Formation", from J. Larossa and 
: U. Cortes. I looked at some of its references ( Peirce, Hobes, 
: Zadrozny...).  
: Nevertheless I am still looking for good definition of 
: abduction, induction and deduction.  I have the feeling that 
: the same process could be called either abduction or induction 
: or deduction, depending of the  context (what we know about 
: the "world"). If it is true, the three could be simulate with 
: the help of the same model (e.g. spreading activation, and 
: levels of activity interpreted as degrees of confidence).

: But is it true ? That is to say, is there some examples in 
: which we can't  transform abduction=>induction=>deduction, 
: more or less progressively, just by adding some pieces of 
: evidence in the known world ?

: Let me try to give an example (adapted from the paper quoted 
: above)
: P1:  Birds(x)=>Flies(x)  (all birds can fly)
: P2: Has_wings(x)=>Flies(x) (anything having wings can fly)

: Then  A:= Bird(x)=>Has_wings(x) (every bird has wings) is 
: called "abduction".

: Now, add P3: "some birds have wings". Couldn't  A be called 
: "induction" ?
: Add P4: "only birds has wings". A should become more plausible 
: (but there might be some birds flying without wings)
: Add P5: "there are  N flying objects and N birds in the 
: world".  Now A is a deduction  
:
*** The notions of abduction, induction and deduction are very different
although they are inter related. You may find an in detail discussion of
these issues in Zeleznikow, J. and Hunter, D., 1994, Building Intelligent
Legal Information Systems, Kluwer and a paper soon to appear in Artificial
Intelligence Review. Inductive reasoning is the process of using statistical
measures to derive rules from cases. The cases are then used deductively.
Deduction is that inference process that states that if certain premises are
true then certain conclusions must follow (if a1 and a2 and ..... an then b).
Abductive reasoning states that from P -> Q and Q it is possible to infer O.
It is an unsound rule of inference.

	John Zeleznikow
  
: Maurice
: http://www.calvacom.fr/calvaweb/Maurice_Clerc/Maurice_Clerc.ht
: ml


