[prev in list] [next in list] [prev in thread] [next in thread] 

List:       aspell-user
Subject:    RE: [Aspell-user] Aspell quality measurement (was: Arabic)
From:       "Binkley, Peter" <Peter.Binkley () ualberta ! ca>
Date:       2006-04-18 15:54:15
Message-ID: 908893006339C0409519E4065DF3B249015711A5 () mailserver ! ualibrary ! ualberta ! ca
[Download RAW message or body]

> Common typos and scannos are not very useful stats since that is not
what soundslike 
> are used for.  If those are the only mistakes soundslike are not
needed at all.  
> What I need are true spelling errors where the user really doesn't
know how to spell
> the word.

In an OCR context, though, it would be handy to have a "lookslike"
instead of a "soundslike". I asked about this a couple of years ago
(http://lists.gnu.org/archive/html/aspell-user/2002-07/msg00002.html),
but I don't have the C smarts to try the solutions that Kevin suggested
then (modifying the weights in TypoEditDistanceWeightsin
util/typo_editdist.hh). It sounds like Lars has the data to attempt this
kind of tuning; if anyone tackles this, I hope they'll announce it here!

Peter

Peter Binkley
Digital Initiatives Technology Librarian
Information Technology Services
4-30 Cameron Library
University of Alberta Libraries
Edmonton, Alberta
Canada T6G 2J8
Phone: (780) 492-3743
Fax: (780) 492-9243
e-mail: peter.binkley@ualberta.ca




[prev in list] [next in list] [prev in thread] [next in thread]