Question about registration

[David Relson]

Hello overlord_q,

On Wed, 25 Feb 2004 01:22:42 -0600
overlord_q at hotmail.com wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> bogofilter version 0.16.4
> ~    Database: BerkeleyDB (3.2.9)

...[snip]...

Versions of both bogofilter and BerkeleyDB are fine.  Nothing to change.

> this is run through a crontab and/or manually through a shell. I get a
> newsletter from a site, and no matter how many times try to unregister
> it is spam and/or register it as ham bogofilter always leaves this
> little tidbit:
> 
> - --------------
> X-Bogosity: Yes, tests=bogofilter, spamicity=1.000000, version=0.16.4
> - --------------
> 
> Any help would be greatly appreciated.

The best thing to do is learn more about which tokens are causing the
message to score as spam.   Have you tried using "-vv" to generate a
histogram or "-vvv" to generate the detailed list of tokens?  They're
described in the FAQ.  You may see some words that rank highly as spam
that are from messages that were classified wrong during training.

To ask the dumb question, you are using "-Sn" to unregister as ham and
register as spam, aren't you?

Also I see that you're using "-u".  Bogofilter does the best it can, but
mistakes do happen and when they do, "-u" will add the incorrectly
classified message to the wordlist.  Do you conscientiously check what's
classified and correct mistakes with "-Ns" or "-Sn" (as appropriate)?

It sounds like the newsletters are significantly different from your ham
and significantly similar to your spam.  If the similarities are strong
enough, bogofilter may be unable to get them right.  With maildrop you
can always take the easy way out and whitelist the mailing list.

HTH,

David