This is turning into a bit of a failboat. Many of the people I asked
would like to help but three key problems stop them:
* They don't trust anyone else with their ham so they wont upload their
corpora anywhere.
* The docs to run masscheck yourself are very poorly written and
confusing. When I personally got started on this I had to ask a ton of
questions here on the list to be sure I was doing the right thing.
* All of their mail is on a remote server. (Syncing this could be done,
but there isn't a good solution for repeated syncing later that
automatically removes mail that you subsequently deleted at the remote
source.)
Let's process the uploaded corpora and see how well the generated scores
do. It will probably be better than the ancient 3.2.x with all these
bug fixes? Just get it out the door, then focus on cleaning up the
documentation/tools and recruiting a greater variety of corpus or
masscheck participants.
If we improve the masscheck sample size substantially, we could safely
redo the scores entirely for 3.3.1.
Warren Togami
[email protected]
