http://bugzilla.spamassassin.org/show_bug.cgi?id=2978
------- Additional Comments From [EMAIL PROTECTED] 2004-01-30 12:14 -------
> Could you try it on some ham as well? All your false positives seem to be on
> stuff that's spam, which isn't really all that interesting, IMHO :-)
Actually all the false positives are indeed from my "ham" folder. Everything
listed is either personal mail or confirmed opt-in, stuff I certainly don't want
tagged as spam.
> (But yes, some of the checks could be removed, replaced by other checks, and
> or
> have their thresholds adjusted.)
I think with some tweaking this would be very useful. I've been toying with
some alternate approaches to this splleling problem, such as "/v[iagr]{4,6}a/".
Also Scott Crosby and I have been working on testing his "Eurika" post from a
few months back where new (unseen) tokens are given a variable bonus in the
bayes engine; initial results look interesting...
> BTW, I'm not familiar with your format, would you please also post some column
> headings? :-)
Standard hit-frequencies output:
OVERALL SPAM HAM S/O SCORE NAME
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
