[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Notice: in file [ROOT]/includes/session.php on line 2208: Array to string conversion
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4688: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4690: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4691: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
[phpBB Debug] PHP Warning: in file [ROOT]/includes/functions.php on line 4692: Cannot modify header information - headers already sent by (output started at [ROOT]/includes/functions.php:3823)
Poco Forums • View topic - More PocoMail Junk Filtering vs. K9
Page 3 of 3

PostPosted: Thu Aug 12, 2004 2:23 am
by vamp07
I agree that PF works very well. The problem is that PocoMail is currently suffering from giving me too many false negatives. Sicne I have a much larger bad words corpus then good words corpus this formula would cause even more false negatives. Maybe I have my logic screwed up and the outcome would be the reverse. Obviously pofile does something right.

PostPosted: Thu Aug 12, 2004 2:34 am
by SFCurley
I think I saw that Slaven is making some changes to the BF process in the upcoming beta. Be interesting to see what effect that has on the false negatives.

As an aside, I'm currently at about 4% false negative rate. I just re-produced my corpus last week, and have had 0% false positives and have improved from 14% false negatives right after I re-produced it to 4% in the last 5 days or so.

PostPosted: Thu Aug 12, 2004 3:54 am
by Omega
vamp07 wrote:I agree that PF works very well. The problem is that PocoMail is currently suffering from giving me too many false negatives. Sicne I have a much larger bad words corpus then good words corpus this formula would cause even more false negatives. Maybe I have my logic screwed up and the outcome would be the reverse. Obviously pofile does something right.


This may be a good reason for you to consider tweaking your Good Word Bias. Regarding the logic question you posted earlier, the probabilities are normalized, so the denominator does not really matter.