[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 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 - Filtering experience

Filtering experience

Discussion on Bayesian and standard junk mail filters

Moderators: Eric, Tomas, robin, Michael

Filtering experience

Postby djgtram » Thu Nov 03, 2005 6:52 am

I'm wondering how other people are using the junk filter in PocoMail, especially seeing other topics here complaining about its performance.

My own experience is that I can get very good results by following a seemingly strange strategy: by allowing the bad word list to grow but always keeping the good list at the minimum. In my experience, allowing the good list to grow will reduce PocoMail's filtering reliability because there will always be words used in both legitimate and spam messages, and those appearing in the good list will do more harm than good. Before actually figuring this out, I had constant problems with messages classified the wrong way. Right now, with 11,290 bad words and 1,037 good ones, accuracy is 99.17% and 0.60% (with 100-200 incoming messages per day). And indeed, it works quite well, I seldom need to overrule its decision. And add the fact that I have regular correspondence in four languages, so PocoMail had to cope with more junk words than usual.

But this also means that, in order to declassify a junked mail without allowing PocoMail to grow its good list, I have to disable junk filtering temporarily. It would be nice to have a way to move a message from the Junk mailbox (via scripting, possibly) without allowing PocoMail to process it. Also, I'd be glad if the 1,000 good word limit could be lifted so that I could operate with no good list at all.

Bye,
Gábor
djgtram
Resident Poster
 
Posts: 100
Joined: Thu Oct 20, 2005 4:51 am
Location: Hungary

Postby mrQQ » Tue Nov 08, 2005 4:22 am

so just reduce good mail bias :0
mrQQ
Frequent Visitor
 
Posts: 66
Joined: Wed Feb 09, 2005 6:03 am

Postby djgtram » Tue Nov 08, 2005 5:37 am

mrQQ wrote:so just reduce good mail bias :0


1 is the smallest setting the dialog box allows. Will this setting ignore good words completely or only reduce their contribution to some extent?

Bye,
Gábor
djgtram
Resident Poster
 
Posts: 100
Joined: Thu Oct 20, 2005 4:51 am
Location: Hungary

Postby mrQQ » Tue Nov 08, 2005 5:53 am

With this setting each good word will be worth as much as each bad word. So if mail has one good one bad word it'll remain neutral (more or less).

at least that's how i think it works
mrQQ
Frequent Visitor
 
Posts: 66
Joined: Wed Feb 09, 2005 6:03 am

Postby djgtram » Tue Nov 08, 2005 6:17 am

Then it won't really help me, I don't want good words to neutralize bad ones, I want them to do nothing at all. Anyway, I already know what I'll do, I'll write a script to be run whenever I start PocoMail which kills the good list by filling it with 1,000 irrelevant words. Then I'll see how this works in real life...

Thanks,
Gábor
djgtram
Resident Poster
 
Posts: 100
Joined: Thu Oct 20, 2005 4:51 am
Location: Hungary

Postby mrQQ » Tue Nov 08, 2005 6:44 am

I don't really see how can you live without good word list - i'd get loads and loads of false positives that way. I currently have 40k bad words, 25k good words (only cause i get way more junk mail than good mail=) and it's working just fine.
mrQQ
Frequent Visitor
 
Posts: 66
Joined: Wed Feb 09, 2005 6:03 am

Postby djgtram » Tue Nov 08, 2005 7:18 am

mrQQ wrote:I don't really see how can you live without good word list.


I always lived without good word lists, I came to PocoMail from Eudora which, as far as I can remember (although, having removed it from my system, I can't actually check now), doesn't use good word lists, either. Junk filtering accuracy was always quite good even with Eudora (above 98%) and it's even better with PocoMail, but only as I do it now, without good words. 99.23% accuracy right now, very few false positives and most of them are a legitimate mistake of PocoMail: empty or nearly empty messages with an attachment, a red flag for any junk filtering but some of those are legitimate for me, I receive some work material in those messages.

I wrote the script I mentioned, now I always have a completely bogus good word list, I'll see how it turns out in practice.

Bye,
Gábor
djgtram
Resident Poster
 
Posts: 100
Joined: Thu Oct 20, 2005 4:51 am
Location: Hungary


Return to Junk Mail Filtering Help and How-To

Who is online

Users browsing this forum: No registered users and 1 guest

cron