> 1. Is Classifier.*_spamprob() thread-safe? I will be calling them  
> from multiple threads. Anything I should be aware of?

1.  You'll want to be calling spamprob() itself, not one of the variants.

2.  As long as you are not *training* from multiple threads (or training and
classifying at the same time), you should be fine.  (i.e. stick to reading
only).

> 2. I've looked around and tried to figure out what would be saved  
> into SB's database. Am I correct in saying that it only saves the  
> classification results and IDs, not the raw message source?

The 'message info' database saves various things, but you won't need any of
those if you're not any of the existing SpamBayes applications.  The hammie
(token) database saves:

  1.  The total number of ham and spam trained.
  2.  (token, hamcount, spamcount) entries.

The 'hamcount' and 'spamcount' in the latter are the number of times that
particular token has been seen in ham/spam.   The raw message source is not
stored.

IMO, it would be best to use the existing SpamBayes infrastructure to manage
the SpamBayes database (writing a new storage class if necessary).

[...]
> 3. At least as the first cut, I'll provide a very dumb-down 
> UI for SB (i.e., no user configuration UI will be provided).
[...]
> Any other default settings I should be aware of? Or using the  
> defaults in Options.py is all I need to do?

All defaults will be fine.

=Tony.Meyer

_______________________________________________
spambayes-dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/spambayes-dev

Reply via email to