Skip to contentSkip to conversation

Bogofilter at VGER..

DISCUSSION Matti Aarnio 3 participants 4 messages Sep 1, 2006 – Sep 4, 2006

lore
Share & tools
lore
#0 lore raw

Hello,

We are considering of taking Bogofilter into use at VGER. So far we are using it in TEST mode - to teach it about SPAM and HAM.

I have added some new "cute" email addresses to VGER to receive any spams that spammers wish to send to us.. See the bottom link at vger's web front page.

You can feed SPAM to [email protected], but do not feed there any HAM. Not that it would really affect statistics in any effective way.

IF we take it into use, it will start rejecting messages at SMTP input phase, so if it rejects legitimate message, you should get a bounce from your email provider's system. (Or from zeus.kernel.org, which is vger's backup MX.)

In such case, send the bounce with some explanations to <[email protected]> -- emails to that address are explicitely excluded from all filtering!

Regards,
Matti Aarnio -- one of <[email protected]>

#1 lore raw

Bogofilter at VGER.. (part 2)

Quoted 5 lines · Matti Aarnio

Hello,

We are considering of taking Bogofilter into use at VGER. So far we are using it in TEST mode - to teach it about SPAM and HAM.

Now after some 30 hour message flow for training the filter has been taken into active use.

It is now able to REJECT things it considers SPAM.

That rejection happens at SMTP input phase, and if your LEGITIMATE email is now rejected (bounced back to you), do complain to VGER's POSTMASTER address ([email protected])
That one address is exempted of all content filterings.

We can _try_to_ train the Bayes to accept your email.

Quoted 3 lines

You can feed SPAM to [email protected], but do not feed there any HAM. Not that it would really affect statistics in any effective way.

My intention was that if you noticed a spam that went through the list, THEN you can feed that back to the above mentioned address. Now several people have fed their entire spam collections there...

Oh uh.. tons of training material..
Please DO NOT DO IT AGAIN.

We have some secret addresses that we use to teach the system about HAM and even to untrain previous SPAM score. We have a number of spam-traps, which I do list below so that they get wider dissemination and end up in spammers address lists... See HTML source of http://vger.kernel.org/bo.html

Regards,
Matti Aarnio -- one of <[email protected]>

Trap addresses:

[email protected] [email protected]
[email protected] [email protected]

  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]       [email protected]
  [email protected]     [email protected]
  [email protected]     [email protected]
  [email protected]       [email protected]

Some of those have two hyphens, some have only one. Copy and paste freely -- especially if you post to usenet newsgroups, but make sure you DO NOT let posts to go to those addresses. (If you e.g. put them into your email cc: headers.)
(Single hyphen ones are based on names I saw in my inbox the other day... Single prefix + somebody's name.)

Signature · 1 line
VGER BF report: U 0.5
#2 lore raw

Re: Bogofilter at VGER.. (part 2)

Quoted 1 line · Matti Aarnio

We can _try_to_ train the Bayes to accept your email.

With the mailing lists on NL.linux.org, I always accept email from list subscribers and use that to train as ham.

Mail to spamtraps is trained as spam, and mail from non-subscribers to the lists is filtered. Through spamassassin though - I need to try out a better bayesian filter...

Signature · 4 lines
What is important?  What you want to be true, or what is true?

-- 
VGER BF report: U 0.5
#3 lore raw

Re: Bogofilter at VGER.. (part 2)

In reply to Matti Aarnio

Hi!

Quoted 8 lines
Quoted 3 lines

We are considering of taking Bogofilter into use at VGER. So far we are using it in TEST mode - to teach it about SPAM and HAM.

Now after some 30 hour message flow for training the filter has been taken into active use.

It is now able to REJECT things it considers SPAM.

Nice, but...

Quoted 1 line

VGER BF report: U 0.5

...

Quoted 4 lines

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html> - Please read the FAQ at http://www.tux.org/lkml/

The list signature is getting long and boooring. Can we move 'vger bf report' to X-bogo-report, and only have one FAQ pointer?

Signature · 4 lines
Thanks for all the (sleeping) penguins.

-- 
VGER BF report: H 0.284667