This is a bit flaky at the moment... sorry. We're working on it.
The sa-learn interface allows you to train SpamAssassin to improve its recognition of spam (unwanted mail) and ham (desirable mail) by feeding it files containing mail messages of one type (spam or ham). After you have authenticated (using either your WatIAM password or the password you use to log in to the mail server you select), you may upload a local (to the machine running your web browser) file containing either spam or ham email (you may need to copy a mail folder to your local disk if you use IMAP to read your mail or locate your mail program's message store if you read your email using POP) and your SpamAssassin settings will automatically be updated to learn from the mail you have uploaded. If you accidentally classify a mail file incorrectly, load it again with the correct classification and SpamAssassin's learning system will be updated. An interface for importing sa-learn data from another mail server is being worked on. At the present time, it only supports thesa-learn --backup file format which requires version 3.0.0
or later of SpamAssassin.