SA Bugzilla – Bug 5822
try URIBL_XBL
Last modified: 2008-04-15 15:49:25 UTC
basically, test URIs against Spamhaus XBL, similar to URIBL_SBL. I'm sure we did this before, but I can't seem to find any measured results! so let's try it again.
: jm 10...; svn commit -m "bug 5822: try out URIBL_XBL (again)" rulesrc/sandbox/jm/20_basic.cf Sending rulesrc/sandbox/jm/20_basic.cf Transmitting file data . Committed revision 620506.
results from last week's network run: http://ruleqa.spamassassin.org/20080308-r634908-n/URIBL_XBL/detail 0.00000 0.0738 0.0056 0.930 0.70 0.00 URIBL_XBL 0.00000 2.2179 0.0261 0.988 0.85 0.00 URIBL_XBL net-bb-jm 0.00000 0.0000 0.0000 0.500 0.40 0.00 URIBL_XBL net-bb-zmi 0.00000 0.0000 0.0296 0.000 0.51 0.00 URIBL_XBL net-cthielen 0.00000 0.0000 0.0000 0.500 0.49 0.00 URIBL_XBL net-dos 0.00000 0.0000 0.0000 0.500 0.47 0.00 URIBL_XBL net-theo 0.00000 6.2315 0.0162 0.997 0.63 0.00 URIBL_XBL net-zmi so about 93% accurate overall, hitting 0.0738% of spam. I think something must be up there, though -- so ignore the submitters apart from bb-jm and zmi, and you get a more realistic 98-99% accuracy, hitting a low 2.2% / 6.2% of spam. I'll try removing #reuse, to see if that improves figures: : jm 574...; svn commit -m "remove #reuse from URIBL_XBL and see if it improves measurement of its accuracy" rulesrc/sandbox/jm/20_basic.cf Sending rulesrc/sandbox/jm/20_basic.cf Transmitting file data . Committed revision 637582.
some results: 0.00000 0.7372 0.0433 0.945 0.83 0.00 URIBL_XBL 0.00000 0.3272 0.0000 1.000 0.81 0.00 URIBL_XBL net-dos 0.00000 0.0000 0.0160 0.000 0.48 0.00 URIBL_XBL net-jm 0.00000 1.8209 0.1489 0.924 0.81 0.00 URIBL_XBL net-theo 0.00000 0.0895 0.0000 1.000 0.52 0.00 URIBL_XBL net-zmi pretty wildly variable! 94.5% accurate overall, with a pretty low hit-rate of only 0.737% spam. I don't think it works out all that well.