Re: fuzzy string matching

Earl Cox (ecox@metus.com)
Mon, 7 Jun 1999 12:45:35 +0200 (MET DST)

I stand admonished!! It's not often that I have my own words thrown back at
me! ;-)

No, it's not necessary that every use of fuzzy sets be part of some larger
machine reasoning system. But two points. First, I went back and read the
original post. I didn't get the impression that the question was about a
spell checker specifically, but rather , about the use of fuzzy logic in
comparing strings in general. Second, I like to always focus on information
gain or the reduction in entropy (ala the ID3 algorithm) in a fuzzy system.
In the case of the fuzzy string operation, I was asking "OK, you know the
degree of difference between strA and strB is [.84] -- what does this tell
you? How is it used in a model of some intelligent process?" In other words,
is it more than an academic exercise?

In fact I went back to Binstock and Rex's really terrific book "Practical
Algorithms for Programmers" and looked at Chapter 4, "Searching". They have
a lengthy section "Approximate String-Matching Techniques" that should a
required jumping off point for anyone who wants to continue this debate.
They also include a detailed discussion of 'Metaphone' an advanced and
powerful extension to the old soundex algorithm for finding strings and
sub-strings that sound alike.

More later.
Earl

Breezy wrote in message <7jdc5d$5ed$1@perki.connect.com.au>...
>Earl,
>
>The original post entitled fuzzy string matching specifically asked for
some
>advise about the use of fuzzy logic in a spell checker application. For
that
>objective I think I am reading your endorsement below for the ability of
>fuzzy sets to generate a metric which could adequately describe error of
>comparison between two strings.
>
>I think I am also reading your frustration for the lack of cognitive value
>bought to the comparison of strings by simply fuzzying of the comparison
>space.
>
>Does every use of fuzzy sets( as a metric) have to be part of a bigger
>generalized machine reasoning system?
>
>Is there not already a good deal of information gain by the matching of the
>semantic (in this case of the mismatch space when comparing the two
strings)
>to the underlying concept (of the misalignment of the comparison) ?
>
>"Fuzzy set theory, and the broader area of fuzzy Logic itself, is not a
>specific method for any particular application any more than Boolean logic
>or probability is a logic for a particular kind of application."
>............The fuzzy systems handbook.
>
>
>Rob Wise.
>
>
>
>Earl Cox wrote in message ...
>>While Lotfi has indicated, in many of his talks, that fuzzy logic is a
>>methodology for adding the calculus of imprecision to any discipline, I
>>still believe that we must look at the informational gain in fuzzy systems
>>rather than simply some method of generating values in the [0,1] interval.
>>Fuzzy logic has become a catch-all phrase for many families of interval
>>arithmetic and for any arithmetic method that generates anything other
than
>>a finite set of integer valued solutions (I recall seeing an investment
>>consultation system that proclaimed it used "FUZZY LOGIC -- the same kind
>of
>>logic found in Mycin and other expert systems!" Of course we know that
>Mycin
>>used a form of semi-adhoc Bayesian evidence calculus (certainty factors)
>>developed by Bruce Buchanan and Ed Shortliffe).
>>
>>etc......
>
>

############################################################################
This message was posted through the fuzzy mailing list.
(1) To subscribe to this mailing list, send a message body of
"SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
(2) To unsubscribe from this mailing list, send a message body of
"UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
to listproc@dbai.tuwien.ac.at
(3) To reach the human who maintains the list, send mail to
fuzzy-owner@dbai.tuwien.ac.at
(4) WWW access and other information on Fuzzy Sets and Logic see
http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
(5) WWW archive: http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html