[Spambayes] Spam Database available?

[KD5NWA]

It was so simple it hurts.

I have a hosting account, I created a new account and told my friends about 
the temporary new address, I then deleted my email accounts, after two 
weeks I created them again and started using them. Six month later, it's 
still pretty clean, I tried before deleting the account for one week, but 
that didn't work.

I think that they basically took me off their list, all I know is that it 
worked, I tried the same thing on my son's email accounts and he is also 
SPAM free so far. I subscribe to about 20 Yahoo groups plus others and so 
far I'm OK.

When I used 1.03 by the time it got about 10 SPAM's it was already figuring 
out a lot of SPAM, but when I switched to 1.04 my problems started.

The problems started when it would classify obvious "SPAM" as "HAM" but 
when you went to correct it, the email was not in the review screen, after 
a short period the database was so messed up it could not figure out 
anything correctly. twice so far I have gone in and deleted all databases 
from SpamBayes and the problem is the same.

With 1.04 I already have over 100 SPAM emails that it has received and 
still all it says is that it's "unsure" it almost never categorizes 
anything as "SPAM" anymore.

I have three PC's that receive the same email, they are not set to delete 
the emails from the ISP unless I delete the email. All three are doing the 
same thing. The funny thing is that I have a older laptop that I don't use 
very much, I upgraded it to V1.04, once a week I fire it up and use it and 
read all the emails that have not been deleted. That PC is running on 
Windows 2000 instead of XP like the others, it's database has not been 
corrupted, and it still extremely accurate at detecting SPAM. I copied it's 
databases from the laptop to the other three PC's and within a couple of 
weeks after constantly making mistakes in classifying mail the database was 
useless.

I'm seriously thinking about going back to 1.03 and hope the problem isn't 
caused by a patch from Microsoft.


At 09:58 AM 6/29/2005, you wrote:
>Training on errors is probably the best way to proceed.  You may be
>surprised how quickly SpamBayes learns.  You could train on old spam if
>you've held onto it, but if you've reduced the volume of spam that
>dramatically, the content of what gets through may be different from
>what used to get through, so such bulk training might not be useful.
>Likewise, training on someone else's corpus of spam is not likely to do
>you much good.  Think of the manual training as the price you pay for
>your phenomenal success in reducing your spam.
>
>Just out of curiosity, how did you knock it down so radically?
>
> > -----Original Message-----
> > From: spambayes-bounces at python.org
> > [mailto:spambayes-bounces at python.org] On Behalf Of KD5NWA
> > Sent: Wednesday, June 29, 2005 9:32 AM
> > To: spambayes at python.org
> > Subject: [Spambayes] Spam Database available?
> >
> > I'm trying to train SpamBayes after the database got
> > compromised, a problem
> > I'm having is right now is that I get about 1 Spam for about 400 Ham
> > emails, at this rate it will take a long time to train the database.
> >
> > I use to get 80% Spam on my email so it wouldn't take long to
> > train the
> > database, but after trying some things to eliminate Spam my
> > Spam rate is
> > down to .025% of the emails I receive.
> >
> > My question  is there available a collection of recent SPAM emails
> > available that I can access to train SpamBayes?
> >
> > Cecil Bayona
> > KD5NWA
> > www.qrpradio.com
> >
> > 'Never argue with an idiot. They drag you down to their level
> > then beat you
> > with experience.'
>
>
>--
>No virus found in this incoming message.
>Checked by AVG Anti-Virus.
>Version: 7.0.323 / Virus Database: 267.8.5/32 - Release Date: 6/27/2005

Cecil Bayona
KD5NWA
www.qrpradio.com

'Never argue with an idiot. They drag you down to their level then beat you 
with experience.'