[Spambayes] Blocked Messages

[Tony Meyer]

>>> Spambayes has blocked some messages from clients. 
>> With appropriate training, false positives should be extremely
>> rare.  It would be worth examining these messages (use the "show
>> clues" feature) to see why they scored what they did. 

Have you tried this?

>>> Is there a way to program Spambayes to send a message 
>>> back to anyone who gets blocked as spam that says 
>>> "you have been blocked as a spam message.  If you think 
>>> this was done in error please call *** ***-****?
>> SpamBayes doesn't actually 'block' any messages at all.  Any messages 
>> classified as spam just end up being filtered into a separate folder.
>> You could simply review this (e.g. once every few days) for any
>> incorrectly classified messages. 
>
> I literally receive hundreds of junk mail messages per day
> so this is not practical for me. 

No spam filter will be able to give you 0 (unrounded) false positives,
without requiring mail to have some sort of compulsory characteristic (to a
certain address, certain word in the subject, etc) - and with such a
characteristic, you invite in higher false negatives (although that may be a
price you are willing to pay).  If you can't manage to have any false
positives at all, then the only choice, no matter how time consuming, is to
manually review all messages.  SpamBayes would at least make this faster,
since it filters off those most likely (based on its training) to be junk.

=Tony.Meyer

-- 
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.