On 2/20/2010 10:39 AM, Jari Fredriksson wrote: > On 20.2.2010 17:29, LuKreme wrote: > >> On 20-Feb-2010, at 08:21, Jari Fredriksson wrote: >> >>> I just found this rule in SA >>> >>> 1.6 FREEMAIL_ENVFROM_END_DIGIT Envelope-from freemail username ends in >>> digit (alv.john2[at]gmail.com) >>> >>> >>> What's the rationale? >>> >> The rational, as with all rules, is that it was run against a massive corpus >> of emails and hit spam far more than ham. >> >> > Ok, fair enough. > > To be more specific, in this case 90.5% of the emails were spam, and 9.5% were nonspam.
This is a sorta-good ratio, but it's not spectacular. Given that the score is only 1.6, and not something larger, that seems to fit pretty well. (There are cases where scores don't line up well with hit ratio, because of interaction with other rules that frequently hit the same messages, but in this case, they match up pretty well. A sorta-good rule with a sorta-good score.)
