[sylpheed:31619] Re: rss-support & spam-filtering

[Bob White]

Hi

I added the -l to the classifying command to get the log message.  So
it's bogofilter -l -I now.

I've never figured out how Bogofilter decides one message is spam and
another is not.  (In principle I think I know but specifically it's a
mystery.)  I've received two identical messages at different email
addresses, one is marked as spam the other is not.  I just assumed
there were minor differences, maybe in the headers, and the score
totalled up a little differently.

Marking some non spam will help a lot.  I put the button on my tool bar
to make it easier.  I don't know of any prebuilt bogofilter files.
Once the filtering kicks in, it does train fairly fast. 

Bob W.



On Mon, 6 Aug 2007 22:05:40 +0200
kazaam <kazaam at oleco.net> wrote:

> Thank you very much for your reply! I also thought that it needed to be trained! But I never marked a mail as non-spam before...
> 
> If I make a right-click on an email and do "mark as spam" it is send to the spam folder I created. But I got the same mail just with a small change in the subjectline, like "Re[12] spamsubject" to "Re[21] spamsubject" (in fact they haven't been answers to anything but just had this "Re" in the subject line) and I marked them both as spam but I still get those same mails just with different numbers behind the "Re" telling me I need a penis enlargment..;) are there any reliable files for bogofilter which are already trained?
> 
> Thats the spam-line: bogofilter -s -I
>       the no-spam line: bogofilter -n -I
>       the classifying spam line: bogofilter -I
> 
> This are my bogofilters settings:
> # bogofilter version 1.1.3
> 
> robx        = 0.520000  # (5.20e-01)
> robs        = 0.017800  # (1.78e-02)
> min_dev     = 0.375000  # (3.75e-01)
> ham_cutoff  = 0.450000  # (4.50e-01)
> spam_cutoff = 0.990000  # (9.90e-01)
> ns_esf      = 1.000000  # (1.00e+00)
> sp_esf      = 1.000000  # (1.00e+00)
> 
> block-on-subnets  = No
> encoding          = utf-8
> charset-default   = iso-8859-1
> replace-nonascii-characters = No
> stats-in-header   = Yes
> thresh-update     = 0.000000
> timestamp         = Yes
> 
> spam-header-name  = X-Bogosity
> spam-subject-tag  = ''
> unsure-subject-tag = ''
> header-format     = %h: %c, tests=bogofilter, spamicity=%p, version=%v
> terse-format      = %1.1c %f
> log-header-format = %h: %c, spamicity=%p, version=%v
> log-update-format = register-%r, %w words, %m messages
> spamicity-tags    = Spam, Ham, Unsure
> spamicity-formats = %0.6f, %0.6f, %0.6f
> 
> 
> On Mon, 6 Aug 2007 13:32:44 -0600
> Bob White <bob at bob-white.com> wrote:
> 
> > I can help with question 2.  First you need to classify some emails
> > using the "Junk" "Not Junk" buttons or in the Message menu.  Second,
> > Bogofilter learns as it goes.  When you get mail in your inbox and
> > classify it as junk using the button or menu, Bogofilter eventually
> > learns which ones are junk.  You do have to specifically mark some as
> > not junk to get started.  Not much seems to happen at first, then it
> > kicks in.
> > 
> > At a terminal, bogofilter -Q will give you the filter settings.  If you
> > have a -l on the bogofilter command line in Sylpheed, a message is sent
> > to the system log each time Bogofilter runs.  You can check there to
> > see what it's doing.
> > [...]
> > Bob W.
> 
>