Hi, I didn't see a response to my original post so I'm posting again... ;-)
I received a spam that got scored only 4.4 yet it had BAYES_80 as well as FORGED_HOTMAIL_RCVD. I'm not very familiar with how much these tags _should_ be scored, but shouldn't a FORGED_HOTMAIL_RCVD have a reasonably high score? And BAYES_80, typically how much does that contribute? I know that high bayes scores will very likely trigger false positives... but I'm wondering how effective the bayes learning is? I know I frequently receive some messages which I push through sa-learn as spam, but they continue to come in with low SA scores... How can I make bayes learning more efficient? Here's the dump from my bayes database: 0.000 0 2 0 non-token data: bayes db version 0.000 0 20179 0 non-token data: nspam 0.000 0 6873 0 non-token data: nham 0.000 0 270474 0 non-token data: ntokens 0.000 0 1064539480 0 non-token data: oldest atime 0.000 0 1065104464 0 non-token data: newest atime 0.000 0 1065090588 0 non-token data: last journal sync atime 0.000 0 1065090603 0 non-token data: last expiry atime 0.000 0 0 0 non-token data: last expire atime delta 0.000 0 0 0 non-token data: last expire reduction count ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
