[Spambayes] Adding a message database

[Tim Stone - Four Stones Expressions]

2/26/2003 7:58:41 AM, Skip Montanaro <skip at pobox.com> wrote:

>    Mark> I would like spambayes to assist in managing a database of
>    Mark> message_ids, mapped to how they were previously trained.  
>    ... much stuff elided ...
>
>I understand what you want to do, but not why.  Can you provide some
>motivation?

Let me speak from recent experience.  I just wrote a Lotus Notes integration, 
and in the process discovered that Notes provides me with almost no facility 
to remember ANYTHING about a message.  It only provides me with a unique 
message id.  I was thus forced to implement a message database, so I could 
remember what had/had not happened to that message.  If I hadn't done that, I 
couldn't have properly retrained a message, and would have had to classify 
every message in the inbox every time I executed the filter.

This pattern was somewhat similar to what happens in the pop3proxy.  Messages 
are given an id, and are managed by that id.  Fortunately in that case, 
information can be embedded in headers.  ***BUT*** headers may not be a good 
place in which to store that information.  Particularly, how the message was 
trained is currently remembered by what subdirectory (Corpus) the message 
lives in.  This idea works for pop3proxy, not for Notes, and not for Outlook.

In Outlook, how a message was trained is currently remembered by what Outlook 
folder the message lives in.  But if I read the continual posts correctly, 
this is an ongoing source of aches, pains, and nausea for Mark, who has to 
handle all of the combinations of user interactions with trained mail to 
correctly untrain/retrain.

Then there's hammiefilter... etc. etc.

All of this adds up (in my mind) to a ton of code rolling around in the system 
simply to manage untrain/retrain which could be nicely abstracted into the 
learn method provided that information about a message could be persisted.

We will undoubtedly encounter more mail systems that could benefit from 
Spambayes.  My Notes work is my most recent example.  I feel that we should 
provide as much facility as we can to make these integrations as easy as 
possible, and this is one such facility.

Thus endeth my apologetic <wink> - TimS


c'est moi - TimS
http://www.fourstonesExpressions.com
http://wecanstopspam.org