|
Training the Bayesian SPAM Filters This page provides instructions for training the Anti-SPAM filters on the college e-mail server. Participation in training the filters is by no means mandatory. Nor do you have to do this every day. If you choose to participate, that's great, and by doing so you will help our filters be more accurate. Message fall into three filter categories:
Please think carefully about what is SPAM and what isn't. If a user receives a message from a major airline about their weekend getaway deals and doesn't want this message, THIS IS NOT SPAM. The user needs to follow the instructions at the bottom of the e-mail to remove themselves from the company's mailing list. This is a message from what could be described as a "legitimate, reputable company" that is likely to remove users from their lists when asked. Please DO NOT label these messages as SPAM. Messages only claiming to be from reputable companies don't count. These usually claim to be from various banks, and want you to give them all kinds of personal information that your real bank should already know. If you get an e-mail from "Dan Druff" about how you can stop hair loss by sending him $3,000, you should definitely NOT respond to the "remove me" instructions, and should label this message as SPAM. Finally, discussion lists: If you want to receive a particular discussion list, make sure you add it to the folder labeled, "I want this discussion list." If you do not wish to participate in the list, unsubscribe from the list- do not mark it as SPAM. Training the Bayesian SPAM Filters, Part II |