Tuesday, November 30, 2004, 7:10:26 PM:

> http://bugzilla.spamassassin.org/show_bug.cgi?id=1050
> [EMAIL PROTECTED] changed:
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>              Status|ASSIGNED                    |RESOLVED
>          Resolution|                            |WONTFIX

> ------- Additional Comments From [EMAIL PROTECTED]  2004-11-30 19:10 -------
> It seems this isn't really an issue... Bayes seems to handle most of these
> situations, and it seems like it would provide little gain.

I disagree. There are different classes of rules, where we can
establish that by the very thing the rule tests for, we're more
confident that rule A identifies spam than does rule B.

Example:  m'[EMAIL PROTECTED]@'/i can be given a much higher confidence level 
for
spam than m'viagra'/i, since normal correspondence could discuss the
drug by name, but only spam will use all available obfuscation
techniques to try to avoid identification.

That said, I have no problem with bug 1050 being closed, since this
confidence issue is adequately discussed in bug 3821.

Bob Menschel



Reply via email to