https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7139
--- Comment #19 from John Hardin <[email protected]> --- (In reply to AXB from comment #18) > (In reply to Kevin A. McGrail from comment #17) > > (In reply to AXB from comment #16) > > > score FSL_HELO_BARE_IP_2 2.000 > > > > > > as I saw S/O in > > > http://ruleqa.spamassassin.org/20150330-r1670013-n/FSL_HELO_BARE_IP_2/detail > > > > > > The axb-8mile corpus is all japanese/chinese spam > > > > I'm confused. Above you said it was bad. This comment implies it is > > hitting on all spam with 100% hit rate. > > its bad S/O in my production but according to my masscheck corpus, > apparently hitting hard on a specific type of spam. Overall masscheck reports an S/O of .996, hitting 33% of spam. That's not too bad. Analysis of current overlaps suggests an additional FP exclusion of !__RP_MATCHES_RCVD would be beneficial (49% of ham, no spam) - would that help your production S/O? -- You are receiving this mail because: You are the assignee for the bug.
