Word in bayes database

[Matt Kettler]

At 10:03 AM 5/24/2004, Nicolas Viers - SCI wrote:
>                  Hello,
>how to remove word(s) in Spamassassin Bayes database ?
>Thanks

You pretty much don't in any easy fashion. If you know how to use the
Berkeley database tools, you can use those, but it's fairly involved and I
don't recommend it if you're not familiar with them. (db_dump and db_load
are the tools)

If you're unfamiliar with those tools, your best bet is to "counter-train".

If the word is declared spam and you really want it to be ham, send
yourself an email with the word in it and force train it as ham.

If the word is declared ham and you really want it to be spam, grab an
inbound spam message and insert the word before training it.

However, I'd strongly caution against tinkering with this kind of thing too
much. Don't be worried about one or two having high spam probabilities in
sa-learn --dump. Only do this if you've got problems with real email.

Of course, deleting the entire bayes database and starting over is an
option if you've got a lot of words (ie: more than 10% of your database)
which misclassified.

-------------------------- MailScanner list ----------------------
To leave, send    leave mailscanner    to jiscmail at jiscmail.ac.uk
Before posting, please see the Most Asked Questions at
http://www.mailscanner.biz/maq/     and the archives at
http://www.jiscmail.ac.uk/lists/mailscanner.html