[spambayes-dev] [ spambayes-Feature Requests-751204 ] Recalculate spam % when you "delete as spam"

[SourceForge.net]

Feature Requests item #751204, was opened at 2003-06-09 02:36
Message generated for change (Comment added) made by swassbac
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=498106&aid=751204&group_id=61702

Category: None
Group: None
Status: Closed
Priority: 5
Submitted By: Dan Boccabella (swassbac)
Assigned to: Nobody/Anonymous (nobody)
Summary: Recalculate spam % when you "delete as spam"

Initial Comment:
I frequently go through my SPAM folder that I have set 
to autmatically filter to and sort by SPAM %.  This way I 
can take a closer look at ones it wasn't as sure about.  
Near the top there are always some that I said delete 
as spam to.  I would think those would be updated to 
100% when I click the delete as spam button.

On a related note, if I say recover as SPAM, why can't 
it set the SPAM % to 0%?

Thanks for a great product!

Dan

----------------------------------------------------------------------

>Comment By: Dan Boccabella (swassbac)
Date: 2003-06-13 13:49

Message:
Logged In: YES 
user_id=797027

I understand that 0%/100% maybe an exaggeration, but just 
recalcing to move the number closer to the right end (so I 
don't still see 0% in the SPAM folder).

Is there any white/black list functionality planned?  Seems 
that I still get an occasional message from a co-worker in 
quarantine - it would be nice to be able to say always (no 
matter what the content) accept mail from this person.

----------------------------------------------------------------------

Comment By: Tony Meyer (anadelonbrin)
Date: 2003-06-12 00:56

Message:
Logged In: YES 
user_id=552329

The "Delete as spam" button does (I believe) rescore the 
message.

The thing is that even if you train a message as ham/spam, 
that doesn't mean that it will score 0%/100%.  It should 
move closer to the correct end, but the score depends on 
how messages have been classified that include the same 
sort of tokens (features) as that message.

For example, if I trained a message as spam, and then trained 
an identical message as ham, the score should be 50% - each 
token is present in one ham message and one spam message.

There's more information (probably clearer!) on the website 
under documentation (start with the FAQ).

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=498106&aid=751204&group_id=61702