bogotune results
Tom Allison
tallison at tacocat.net
Wed Mar 24 12:51:03 CET 2004
Boris 'pi' Piwinger wrote:
>>>Was this before or after running bogotune?
>>
>>After.
>>Before my test results in 100.00% scores.
>
>
> So actually bogotune gave results worse than your orignal
> settings. Interesting.
>
I had trained everything repeatedly on these 4000+ emails to get 100%
scores. I was surprised to see how high ham_cutoff we set. It is equal
to the highest single ham score. I would have expected a nudge up from
there to avoid the 1 fpos.
>
>>Yeah, I guess I can keep bumping up the cutoffs.
>>I set ham_cutoff=0.24 and spam_cutoff 0.4 for now.
>
>
> Why don't you set both to the same. If you get an unsure or
> a false negative, you have to deal with it either way. You
> don't expect any ham in unsure anyway.
>
I guess I could, but I use the unsure range as a buffer.
Currently I'm delivering Unsure mail and using that "bucket" for
additional training on new messages. Because of this, I don't really
mind Unsure ham as I don't count it as a false reading.
I'm going to play with these numbers a bit more in the coming months,
but I'm surprised at the results from bogotune.
More information about the Bogofilter
mailing list