Thanks for the reply. I think I didn't explain myself very well. I guess the question is how do I effectively make the system learn? Will it do it automatically (how, what's the intelligence like), or will I need to provide good and spam emails for it to work on?
For it to work as designed, each user needs to continually submit as much of their spam and legitimate E-mail as possible, and run it through the training program.
Some people have tried to automate this, but it is impossible to do so (to do so automatically requires knowing if the E-mail is spam or legitimate; if you can figure that out automatically, you don't need the Bayesian filtering).
However I'm keen to use the imail analysis to it's full potential. I know you can give it a mailbox to learn from, but I was wondering if I could output a list of spam and good from outlook in to a text file and then call it folder.mbx for the antispamseeder.exe tool to work on.
That is what you should do, but unfortunately I can't say how it would be done (I've never really used IMail v8's anti-spam).
The problem is that if you don't do this for all users, you will find that the statistics do not work properly for some users. For example, for an average user, the word "mortgage" in an E-mail may indicate a 99% chance of spam. However, for someone who works at a bank or someone who is about to buy a house or someone who only recently started getting spam, it may be well under 50%. If one of those people who differs from the norm doesn't train the engine, their legitimate E-mails will be more likely to be marked as spam.
That's why we never incorporated Bayesian filtering in Declude JunkMail. We experimented with it years ago (well before any anti-spam program incorporated it!), but found that the benefits did not outweigh the drawbacks (the constant training by all users, and the problem with spammers adding random words to their E-mails).
Having search the list, some people swear by ASSP. I guess there seems no point in having some detection if you can not update it easily.
That all depends on what angle of attack you use in detecting spam -- for example, Declude JunkMail doesn't require any training, downloading definitions, etc. Instead, it uses techniques that we have developed since 1997 to detect patterns in spam without having to look for key words.
-Scott
---
Declude JunkMail: The advanced anti-spam solution for IMail mailservers since 2000.
Declude Virus: Ultra reliable virus detection and the leader in mailserver vulnerability detection.
Find out what you've been missing: Ask for a free 30-day evaluation.
--- [This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)]
To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/ Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/
