Learn Bayesian filter |
Post Reply ![]() |
Author | |
Alexey ![]() Guest Group ![]() |
![]() ![]() ![]() ![]() ![]() Posted: 29 September 2004 at 3:40am |
Hi. Sorry for my bad English. I have version 2.1.1.367. How I can learn bayesian filter by hand? |
|
![]() |
|
LogSat ![]() Admin Group ![]() ![]() Joined: 25 January 2005 Location: United States Status: Offline Points: 4104 |
![]() ![]() ![]() ![]() ![]() |
Alexey,In theory you could "inject" emails manually in SpamFilter to tweak the Bayesian filter, but that is usually a very bad idea. The Bayesian filter examines statistically the full email, headers included. Your artificial emails would not typically reflect the same content a true spam email will have. The headers may be different, if you copy and paste the content of an email from MS Outlook for example, the original source of the email will be very different since Microsoft Outlook completely changes the source, etc...Roberto F.
LogSat Software
|
|
![]() |
|
Alexey ![]() Guest Group ![]() |
![]() ![]() ![]() ![]() ![]() |
Well, i (and not only me) CAN to feed SpamFilter source email (and i'm not using MS Outlook :-)). You are made Bayesian Probability on Bayesian Filter settings tab, why not to make, for example, check box that say filter that email is evident spam. Supplementary flexibility will only for good :-) By the way, after 5000 spam and 5000 not spam emails, spam email will quarantined or rejected? |
|
![]() |
|
LogSat ![]() Admin Group ![]() ![]() Joined: 25 January 2005 Location: United States Status: Offline Points: 4104 |
![]() ![]() ![]() ![]() ![]() |
Alexey,As I've stated, yes, you can inject emails if you want, but we do not feel it's usually a good idea. Bayesian filters are usually very accurate because the mathematics they are based on assume a real case scenario of text received. If you alter the real life by forcing certain tokens to receive a higher spam score, you risk altering the way a legitimate email is analyzed. If for example a legitimate email contains one of the tokens you artificially modified, the overall spam score assigned to it will not be accurate.If however you feel confident that artificially altering some of the tokens will help the accuracy, you're absolutely free to try. Any email that SpamFilter processes is analyzed by the Bayesian engine. If you wish to inject tokens as spam, you can simply create an artificial email and for example change any of your blacklists so that it is blocked, which will cause the Bayesian filter to categorize the tokens as spam.As far as your last question is concerned, mail blocked by the Bayesian filter will be quarantined if the quarantine database is configured, otherwise it will be rejected.Roberto F.
LogSat Software
|
|
![]() |
|
Alexey ![]() Guest Group ![]() |
![]() ![]() ![]() ![]() ![]() |
Roberto, I see. Why i bother with this questions? By default, SpamFilter comes without pretrained bayesian filter and that why ALL mail are legitime if quarantine does not work. I have to switch on quarantine that bayesian filter start learning. That why need manually learning. But unfortunately i can't make one thing. Larger part of spam mail comes to non-existent users. I can't direct it to bayesian filter. Now i use WhiteTO list. May be you are prompt me how do this. |
|
![]() |
|
LogSat ![]() Admin Group ![]() ![]() Joined: 25 January 2005 Location: United States Status: Offline Points: 4104 |
![]() ![]() ![]() ![]() ![]() |
Alexey,You don not have to turn on quarantine in order for the Bayesian filter to work. You just need to let the other filters (SpamFilter comes with default settings that will already catch a lot of spam) to their job. As emails arrive and are blocked, the bayesian filter will learn and eventually kick in as well.Roberto F.
LogSat Software
|
|
![]() |
|
Alexey ![]() Guest Group ![]() |
![]() ![]() ![]() ![]() ![]() |
Maybe I don't understand something? For example, I'm installing SpamFilter on new system. Quarantine is off, bayesian corpus have zero size. HOW bayesian filter can distinguish spam from non-spam mail?
|
|
![]() |
|
LogSat ![]() Admin Group ![]() ![]() Joined: 25 January 2005 Location: United States Status: Offline Points: 4104 |
![]() ![]() ![]() ![]() ![]() |
SpamFilter has several other filters that block spam. For example the MAPS servers, the "No Reverse DNS" server, the keyword filters. All of these will block spam, even if the quarantine is disabled. Please also note that all the quarantine does is to save a copy of blocked emails in a database so they can be retrieved later.Roberto F.
LogSat Software
|
|
![]() |
Post Reply ![]() |
|
Tweet
|
Forum Jump | Forum Permissions ![]() You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |
This page was generated in 0.374 seconds.