I think this idea has a lot of merit, but I have the feeling that we might
run into a number of unforseen problems; however, partial word substitutions
might help a lot. The book I am presently validating wants to substitute di
for th. I have found that find and replace works best to solve this problem;
however, in one case the correct word was "die" which was read as "the."
----- Original Message -----
From: "E." <thoth93@xxxxxxxxxxxxx>
To: <bksvol-discuss@xxxxxxxxxxxxx>
Sent: Friday, August 19, 2005 4:31 AM
Subject: [bksvol-discuss] we borrow the earth
I am validating this fascinating book. It started out with 96% of the words spelled correctly. I have it up to just under 99% with a lot of assistance from rank spelling.
I have a suggestion for a rank spelling feature or add on. When I notice a pattern in words where a whole group are listed as rank spelling errors I could input a letter substitution to the utility and try out the results in a "read word in context box" before finally accepting the letter substitution. In this book v is often scanned as u resulting in inuite uisit uent preuent auailable and so on. Use of a tool such as this might also include the ability to bring together adjacent miss-recognized word fragments as in "part icularly". Again we would want a context checker box before final acceptance of the letter substitutions or space removal for word fragments.
Also, we really should put a floor on what is submittable or put a strange responsibility on the step 2 validators. We are getting back to what to do with truly poorly scanned stuff here. Could there be some kind of screening before all scans are put on the step 1 page or is that just too politically incorrect?
E.
-- No virus found in this incoming message. Checked by AVG Anti-Virus. Version: 7.0.338 / Virus Database: 267.10.13/78 - Release Date: 8/19/2005
