On Tue, 28 Nov 2006, Evan Platt wrote:

> Thanks to everyone especially Decoder, I think I'm up and running.
>
> png is the only one not working.
>
> Any reason NOT to assign 10 points to fuzzy ocr tripped words?
>
> I mean I wouldn't add 10 points just because someone typed the V word
> in an e-mail to me, but I can't think of an instance where I'd expect
> a GIF message with it in it.

You -do- understand that the 'fuzzy' part of FuzzyOCR means that it
does inexact matching on the characters that it pulls out of an
image. So for example, a college newsletter that I received which
had a school logo image fired on FuzzyOCR claiming to match "company".

I've also seen it fire on things such as an airline ticket confirmation
notice, a religious newsletter, and a technical bulletin. Just one
word for each, which with the default score wasn't enough to tag
as spam but with a score of 10 a guaranteed FP.

Dave

-- 
Dave Funk                                  University of Iowa
<dbfunk (at) engineering.uiowa.edu>        College of Engineering
319/335-5751   FAX: 319/384-0549           1256 Seamans Center
Sys_admin/Postmaster/cell_admin            Iowa City, IA 52242-1527
#include <std_disclaimer.h>
Better is not better, 'standard' is better. B{

Reply via email to