Testing shows katastrophy
Boris 'pi' Piwinger
3.14 at logic.univie.ac.at
Wed Jan 22 16:43:48 CET 2003
Boris 'pi' Piwinger wrote:
>> [3.14 at pi ~]$ bogoutil -w ~/.bogofilter .MSG_COUNT
>> spam good
>> .MSG_COUNT 14 5410
>>
>> Looks really bad. So does this mean I should split with
>> formail when training?
>
> I did this:
>
> [3.14 at pi ~]$ grep -v '^#' .bogofilter.cf |grep .
> algorithm=fisher
> min_dev=0.1
> ham_cutoff = 0.00
> header_format = %h: %c, spamicity=%p, version=%v/%a
> [3.14 at pi ~]$ bogoutil -w ~/.bogofilter .MSG_COUNT
> spam good
> .MSG_COUNT 4186 15000
>
> Looks much better
Test is still runnig, but I gotta go, so here you get the
results I have so far:
Spam:
4186 test.spam
False negatives:
364
Ham:
So I am not too happy with the false negatives. What
parameters can I change without having to rebuild the database?
My understanding is that tweaking min_dev and spam_cutoff
would be OK, right? How about changing the agorithm to robinson?
pi
More information about the Bogofilter
mailing list