case folding [was: tuning ]
David Relson
relson at osagesoftware.com
Thu May 8 18:39:09 CEST 2003
At 12:32 PM 5/8/03, you wrote:
>The spambayes filter (spambayes.sourceforge.net) converts everything
>to lowercase except in the subject line. According to their tests,
>the subject line is the only place where case made a significant difference.
>
>Most of their testing, however, was done with databases of up to 1000
>emails in each category.
Bob,
Bogofilter has an option to create special tokens (with a "subj:" prefix)
from words in the Subject: line. It's likely a small change to skip case
folding for those tokens.
David
More information about the Bogofilter
mailing list