wordlist.db problem
OTR Comm
otrcomm at isp-systems.com
Fri Jun 18 06:45:37 CEST 2004
Hello,
David Relson wrote:
> You're the first to report such a problem with 0.90. Are you sure
> you've got the proper path for your wordlist file and that it has proper
> permissions?
I don't think the problem is with bogofilter, I think the problem is
that I don't have a large enough corpus in my database. That is, I
discovered that if I don't use the -u switch to automatically update the
database when I test for spamicity and let bogofilter work with what is
in the database, it makes a best guess based on what it knows. Then
after I correct the database on an error, and go back at it with:
bogofilter -e -p -d ./.bogofilter < spam/spam.31514,
it works fine.
I don't want to automatically update the database anyway, so this
problem made me dig a little deeper into how bogofilter works. Still a
long way to go, but this has been helpful. I may be wrong about not
using the -u switch, but I don't see that it buys me much. Or does it?
>
> Evidently you're using .bogofilter as your directory name. The usual
> directory is ~/.bogofilter, which can also be written as
> $HOME/.bogofilter. However, .bogofilter is also acceptable, so long as
> you're running bogofilter from the proper directory.
>
> What is the output of command "ls -l .bogofilter/wordlist.db"?
-rw-r--r-- 1 otrcomm otrcomm 16384 Jun 17 19:24 wordlist.db
>
> Bogofilter has some debugging switches which will make it print info
> about database operations. Try adding switches "-x d -vv" to your
> command, as in "bogofilter -d .bogofilter -s -x d -vv < spam/spam.31514"
> and post the output.
[pid 28247] DB->open(db=0x806c0e0, file=.bogofilter/wordlist.db,
database=NIL, type=1, flags=0=, mode=0664) -> 0 Successful return: 0
# 139 words, 1 message
db_close (.bogofilter/wordlist.db) sync
>
> Also you can use bogoutil to display the contents of the wordlist, as in
> "bogoutil -d .bogofilter/wordlist.db". If you're interested in the spam
> and ham counts for particular words, run "bogoutil -w
> .bogofilter/wordlist.db word1 word2 word3".
>
> In your example, the score of 0.520000 is what I'd expect when
> bogofilter is running with an empty wordlist (or has been given the
> wrong location for the wordlist).
I think I figured something else out. Not sure, but when I was
originally testing bogofilter, I was logged in as 'root', but when I
logged in as 'otrcomm' and ran the tests you suggested above, everything
worked fine. However, I still don't understand the value of the -u
switch.
One other question though, how does bogofilter ever come up with an
'Unsure' classification? It always classifies mine as either Yes or
No? I thought that it would have some bound around .5 probabliity that
would trigger an 'Unsure' classification. Is this somewhere in
bogofilter.cf.example that I missed?
Thanks,
Murrah Boswell
>
> HTH,
>
> David
> _______________________________________________
> Bogofilter mailing list
> Bogofilter at bogofilter.org
> http://www.bogofilter.org/mailman/listinfo/bogofilter
More information about the Bogofilter
mailing list