http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5768
------- Additional Comments From [EMAIL PROTECTED] 2008-02-12 14:47 ------- (In reply to comment #9) > For testing DOB please understand the following...the hit rates will change depending on the day. The > DOB lists domains that are less than 6 days old on com/net/biz/org/info/us > > If you test it with URIs in the spam that are using domains/ip that are related to domains < 6 days > old you will have a completely different hit rate than testing against domains that are > than 5 > days old. If I look over the messages in my spam box for the past 2 weeks (from 20080201 to today), and checking the rules they hit when they arrived, I get 39 hits on my spam for RCVD_IN_DOB, out of 48198 messages! that's a 0.08% hitrate, which correlates with comment #4. so that rule is not much use. The same is true for DNS_FROM_DOB; 39 hits, and I suspect they're the same messages (although didn't bother checking). however: URIBL_RHS_DOB hits 13936 out of 48198. that's pretty good, 28%. I've just noticed -- it was not set with #reuse. THis is now fixed, so the accuracy rates in ruleqa.spamassassin.org should start working for that rule: : jm 180...; svn commit -m "oops. URIBL_RHS_DOB and URIBL_RHS_AHBL need #reuse so that their accuracy is measured correctly" rulesrc/sandbox/ Sending rulesrc/sandbox/jm/20_dob.cf Sending rulesrc/sandbox/jm/22_bug_5667.cf Transmitting file data .. Committed revision 627159. I've removed the other rules from trunk, since they aren't hitting enough to be worthwhile: : jm 194...; svn commit -m "bug 5768: actually, remove DNS_FROM_DOB, RCVD_IN_DOB and __RCVD_IN_DOB entirely, due to low hitrates" rulesrc/sandbox/jm/ Sending rulesrc/sandbox/jm/20_dob.cf Transmitting file data . Committed revision 627162. ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
