-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Robin Lynn Frank wrote:
|>2) Spamassassin (with vipul and bayes) scores email | | Vipul is a matter of choice. We don't use it because we haven't found it | better than other resources.
I usually try and use multiple independant indicators to help reduce false positives, SA, VIPUL, and BAYES all use different methods so they tend to compensate for each other. If there were a fourth method I would use that as well.
|>3) High scoring messages are challenged | | Assuming that high scoring messages are spam and therefore likely to have | forged headers, why challenge these?
Because I've got my SA threshold set at 3.0 This score has the lowest combined rate of false positives and negatives according to the included statistics, but if we just quarantined we'd have people missing messages fairly regularly. I'm looking into doing something different with REALLY high scoring messages like over 10.0, but I haven't decided if it's worth it as of yet.
| If not, and if it is not a mail list or trusted sender, it goes to SA. | High scores get quarantined. Everything else goes to TMDA. (At this point, | there is very little that gets challenged.)
I'm trying to keep down the number of challenges to legimate but not whitelisted senders so as to lessen TMDA's visible negative impact.
- -- Chris Berry [EMAIL PROTECTED] Systems Administrator JM Associates & Coast Business Service
"Some days you fix the multi-million dollar machine, other days the $12 stapler kicks your ass." -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFAIWBakAS13ByzgbsRAsImAJ0YL4uNu9L042fRgdpz459ypRWpNwCghoKf vQn76sxeY282IqxoLJgOsQw= =caxV -----END PGP SIGNATURE----- _____________________________________________ tmda-users mailing list ([EMAIL PROTECTED]) http://tmda.net/lists/listinfo/tmda-users
