[Spambayes] cant use this program

Tony Meyer tameyer at ihug.co.nz
Thu Jul 29 06:41:05 CEST 2004


> What im guessing i do is click on the icon in my task bar
> and when the page comes up, train the message as spam.

Yes, every now and then (probably each time you check your mail at first,
but pretty soon every few days should be sufficient) you'll want to train
the system on both ham (good mail) and spam (junk mail).

> I don't
> understand what the instructions meant by dbx or mbox file? 

A dbx file is what Outlook Express stores it's messages in.  A mbox file is
what most other mail programs store their messages in (or close enough).  If
you want to train a whole lot of messages at once, you can use this: put
them in a folder, then select the folder's dbx file.  If you're just
training individual messages, which is usually better, then you can ignore
that.

> It seems like alot of work to train each email.

It's to train a whole lot of messages at once (for example, giving the
system some initial training, based on good and bad mail that you happen to
have around).  There's no need to do it for single messages.

> Can i just save a pile of deleted ones and somehow get it someplace 
> where i can tell the program its all spam?

This is the idea - put them all in a folder, and then select the dbx file by
that name.

> This isnt a program that just works i gather. it seems like for it to
work,
> i have to do alot of stuff , is that accurate.

You have to train it, yes - that's what makes it good.  There are spam
filters that don't need training, but these typically have a fixed rule set.
That means that someone else decides what is good mail and what is bad.
With SpamBayes, *you* train it to recognise what *you* think is good and
what *you* think is bad.  If you like offers for fake diplomas, and hate
mail from Uncle Joe, then you can train it like that - SpamBayes won't care.

However, SpamBayes learns pretty quickly.  You shouldn't have to train it
for long before it's very accurate.  Probably the best training method for
you is to just train on anything that SpamBayes gets wrong - i.e. any mail
that it is unsure about, and any ham it calls spam and spam it calls ham.
Pretty quickly there won't be much of that, so you won't have to do much
training.

It sounds like you're not aware of the "Review" page, either.  There's a
link to this from the main page, plus one from the tray application.  From
here, you get presented with a list of all the messages that SpamBayes has
classified recently, and you indicate what training you would like to do.
You can set the defaults to whatever you like, so training should be a very
quick exercise.

=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. This
way, you get everyone's help, and avoid a lack of replies when I'm busy.



More information about the Spambayes mailing list