Performance of Bogofilter, etc.

[Boris 'pi' Piwinger]

Jesse Trucks wrote:

Please don't send personal copies.

> I started with about 500 spam messages I saved from my own spam, then
> ran my ~33,000 saved ham messages

OK, that sounds dangerous. I guess you should try to have
roughly (give or tage a magnitude;-) the same amount of
messages. Or you use a script which does not really care
about the input size like my training script.

> (all my own mail going back a few
> years) and the tiny bit of spam. The obvious result was a lot of false
> negatives, but after 4 months, I have over 14,000 of my own spam saved.
> I only get about 5 or 6 false negatives (spam in my inbox) on an average
> day now,

Still a log. Have you tried bogotune?

pi