[svlug] procmail stochastic spam filter

John Conover conover at rahul.net
Wed Jul 18 10:59:01 PDT 2001

Hi Jeff.

It actually does a correlation between properties.

The conditions, "* 12345^0 somecondition" are a correlation-12345 is
-10000 * ln (frequency of occurrence of somecondition), so when they
are added, it performs a multiply, and gives the probability that the
various conditions correlate, (actually, its a math trick, so that it
is done in positive integer arithmetic-that's the reason for the
multiply by -10000.)


Jeffrey Siegal writes:
> John,
> Maybe there's another list to discuss this (I don't know), but wouldn't
> you get significantly better discrimination if you took into account the
> correlation between properties?  Or maybe that's just too hard to do in
> procmail?
> _______________________________________________
> svlug mailing list
> svlug at lists.svlug.org
> http://lists.svlug.org/lists/listinfo/svlug

John Conover        Tel. 408.370.2688  conover at rahul.net
631 Lamont Ct.      Fax. 408.379.9602  http://www.johncon.com/
Campbell, CA 95008  Cel. 408.772.7733  

More information about the svlug mailing list