Training problems on version 1.0.2

[David Relson]

On Mon, 24 Jul 2006 18:45:54 +0200
Nigel Henry wrote:

> This is my first experience with spam filtering, so freely admit that
> I don't know what I'm doing.
> 
> Someone on the the KDE list, as I'd asked about Kmail and spam
> filtering, suggested Bogofilter, so here I am. 
> 
> I have installed Bogofilter from the 1.0.2 tarball, which has
> installed it in /usr/local/bin .  It's also put the example config
> file in /etc , and I havn't messed with that, as a guy who wrote a
> tutorial said he'd left the config file as it was, and had Bogofilter
> working ok. I created a directory, .bogofilter in my home directory,
> and 2 new directories in Kmail, named, Spam, and NotSpam, and have
> been adding emails to both of them.
> 
> Now I'm trying to train it. As user, I'm cd'ing to Mail/Spam/cur, and
> then running, bogofilter -sv, which creates a 8KB wordlist.db
> in .bogofilter.  On the Konsole nothing happens, no verbose output,
> no return to the command prompt after about 30mins, and I have to
> quit the shell from the session menu. Running the following, shows
> whats in .bogofilter/wordlist.db.
> 
> [djmons at localhost djmons]$ bogoutil -w ~/.bogofilter .MSG_COUNT
>                                  spam   good
> .MSG_COUNT                          0      0
> [djmons at localhost djmons]$ bogoutil -d ~/.bogofilter/wordlist.db
> .MSG_COUNT 0 0 20060724
> [djmons at localhost djmons]$
> 
> Kmails email directories are in the maildir format.
> 
> Any suggestions will be appreciated.
> 
> Nigel.

Nigel,

8k is the size of an empty DB4 database.  It sounds like bogofilter
hasn't entered any data, as yet.  The multi-minute delay sounds like
bogofilter is waiting patiently for its input.

When you run bogofilter
how are you identifying the input message (directory)? The following
techniques are standard for these tasks:

  bogofilter -sv < message
  bogofilter -sv -B directory

With your directories, initializing the database would use commands:

  bogofilter -sv -B Mail/Spam/cur
  bogofilter -nv -B Mail/NotSpam/cur

Hope this helps!

Regards,

David