Michael> I have been investigating an issue that seems to be pointing at
Michael> Spambayes, but I am not 100% sure yet. When Spambayes 1.04 is
Michael> installed on an Windows XP SP2 system, with Outlook 2007, I am
Michael> getting lots of user reports about delays.... Have you had any
Michael> other reports like this?
No, we don't generally get reports of slowness. However, SpamBayes
performance is dependent on the number and size of mails being classified at
one time the size and type of the training database and to the way it is
used. Do you know any of these parameters for your users?
* Are they using local or network storage?
* Do they tend to start and stop Outlook frequently (thus causing it to
process a bunch of email at startup) or do they leave it running (so
it processes emails as they arrive one-by-one)?
* Are they using a Python pickle file to hold their training database
(read once into memory at startup) or are they using one of the hash
file options (file accessed as SpamBayes runs)?
* How many ham and spam emails are in their training databases? Tens?
Hundreds? Thousands? Some people get carried away and train on every
email they receive resulting in enormous databases. While the
training data are stored in hash files or in Python dictionaries,
there is still a higher cost to looking up words in large databases.
Retraining from scratch and only training on mistakes and unsures
should help keep their database sizes manageable.
--
Skip Montanaro - [EMAIL PROTECTED] - http://www.webfast.com/~skip/
"Be different, express yourself like everyone else."
_______________________________________________
SpamBayes@python.org
http://mail.python.org/mailman/listinfo/spambayes
Info/Unsubscribe: http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html
