Lots of spam gets through because of BAYES_00 -2.60

[Chris W. Parker]

On Tuesday, September 11, 2007 1:19 PM Alex Broens said:

> are you autolearning to Bayes?

I think so. "autolearn=spam" right?

Sep  9 22:37:01 filter MailScanner[18715]: Message l8A5ZuaQ026042 from
76.87.143.41 (berke at pop3.connect.ie) to swatgear.com is spam,
SpamAssassin (not cached, score=26.644, required 4.5, autolearn=spam,
AXB_XMID_1212 3.50, BAYES_99 3.50, BODY_ENHANCEMENT 0.31, DCC_CHECK
2.17, FH_HELO_EQ_D_D_D_D 0.00, HELO_DYNAMIC_DHCP 1.40,
HELO_DYNAMIC_IPADDR 2.43, RCVD_IN_XBL 3.03, RDNS_DYNAMIC 0.10,
SARE_ADULT2 1.42, STOX_REPLY_TYPE 0.00, URIBL_AB_SURBL 1.86, URIBL_BLACK
1.96, URIBL_JP_SURBL 1.50, URIBL_OB_SURBL 1.50, URIBL_SC_SURBL 0.47,
URIBL_WS_SURBL 1.50)

> if yes, are you rejecting so much spam that BAyes doesn't get a chance
> to get enough crud to learn?

I don't know. How can I determine this?

> if not, what are you doing to "teach" it?

Autolearn only. I don't know how to feed messages to it.

> do you realize that Bayes is just a few extra points in a sum of
> rules? 

Yes.



Thanks,
Chris.