Newsgroups: comp.ai.fuzzy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!cs.utexas.edu!swrinde!gatech!newsfeed.pitt.edu!dsinc!ub!acsu.buffalo.edu!dmark
From: dmark@acsu.buffalo.edu (David Mark)
Subject: Discrete domains (was Re: Continuous vs. piecewise-linear MBFs)
Message-ID: <CzFGBr.E25@acsu.buffalo.edu>
Sender: nntp@acsu.buffalo.edu
Nntp-Posting-Host: autarch.acsu.buffalo.edu
Organization: UB
References: <3adrr6$28k@news.tamu.edu> <3aecv4$inq@ixnews1.ix.netcom.com>
Date: Thu, 17 Nov 1994 19:47:51 GMT
Lines: 16

What do fuzzy set folks do if the membership function is over a domain with
a number of discrete states which cannot be ordered along any single
axis to produce a single-valued 'membership function' of the type commonly
discussed?  We are modeling a domain in which a theoretical model distinguishes
exactly 19 discrete 'states'.  Then, we have a set of 60 stimuli, 2 or more
distinct instances for each of the 19 possible states.  Then we showed
subjects these 60 stimuli, and asked them to rate how well a given sentence
fit that stimulus, on a scale of 1 to 5.  From those, we can compute an
empirical estimate of the 'membership' of each stimulus in the concept
represented by the given sentence.  We can also determine the mean response
over all stimuli belonging to each of the 19 states.  But as I said before,
the 19 states cannot be 'ordered'.  What literature is there on applying
fuzzy logic to situations like this?

David Mark
dmark@sun.acsu.buffalo.edu
