Skip Montanaro said:
>
> I've seen several recent references in SA-talk to "S/O" and "S/O Ratio". I
> don't find it mentioned in the 2.30 sources I have laying about and googling
> for it didn't turn up any obvious winners either. All I can think of is
> "Spam to Ollie ratio". What is it really?
Spam-to-Overall Ratio: a Dan Q innovation! ;) it's the ratio of spam hits
to all hits in a given corpus, ie.
nspam
s/o ratio = --------------
(nspam + nham)
(Excuse my use of the "ham" terminology, but I like it ;)
S/O ratio ranges from 0 to 1.0, and values nearer 0 mean more ham-hits,
values nearer 1.0 mean more spam-hits. So 1.0 (for a normal,
non-compensation test) is a perfect 100% spam, 0% false-positive test.
Basically if a mail triggers any given rule, the rule's S/O ratio is the
probability that that the mail is spam. (This means that it's exactly the
same as the P(x) probability values used in Bayesian probability, BTW.
[1])
Recently, it's what we've been using to measure test effectiveness.
([1]: In fact, it could be said that the proposed ranged GA evaluator
effectively uses Bayesian probability with a second, perceptron layer on
top, I think.)
--j.
-------------------------------------------------------
Sponsored by: AMD - Your access to the experts on Hammer Technology!
Open Source & Linux Developers, register now for the AMD Developer
Symposium. Code: EX8664 http://www.developwithamd.com/developerlab
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
