Introduction
Updated: 24/03/2021
Article #: 99
Bayesian filters learn the difference between your good email and spam
email and become very accurate after a brief period of training.
After enabling the Bayesian Filter (Settings >> Filters >> Settings) and setting the amount of email to train, incoming email is collected in the Bayesian screen (Settings >> Filters >> Bayesian Training). On this Bayesian Training screen, once you have the recommended number of email listed, you mark email as good (thumbs up) or spam (thumbs down). Ideally you want around 50% spam and 50% good email.
When you are ready to use the bayesian filters, click the CHANGE button to make the filters active. The incoming email collected to the Bayesian Training screen will stop and the bayesian filters will start classifying your incoming email.
Note 1 - You may wish to disable bounce/undeliverable notifications (Settings >> General) as some false positives are likely at the beginning of bayesian training.
Note 2 - Rescuing an email from Quarantine will add the email to the Bayesian Training screen as a good email and update the bayesian filters.
Note 3 - You may wish to enable Spam Reporting (Settings >> General) as any spam reported will update the bayesian filters. Spam Reporting appends a spam reporting link to all non-whitelisted incoming email.
Note 4 - You may wish to add legitimate email addresses to the whitelist to avoid the bayesian filter entirely. That way the bayesian filter will only need to be trained on outlying legitimate emails.
After enabling the Bayesian Filter (Settings >> Filters >> Settings) and setting the amount of email to train, incoming email is collected in the Bayesian screen (Settings >> Filters >> Bayesian Training). On this Bayesian Training screen, once you have the recommended number of email listed, you mark email as good (thumbs up) or spam (thumbs down). Ideally you want around 50% spam and 50% good email.
When you are ready to use the bayesian filters, click the CHANGE button to make the filters active. The incoming email collected to the Bayesian Training screen will stop and the bayesian filters will start classifying your incoming email.
Note 1 - You may wish to disable bounce/undeliverable notifications (Settings >> General) as some false positives are likely at the beginning of bayesian training.
Note 2 - Rescuing an email from Quarantine will add the email to the Bayesian Training screen as a good email and update the bayesian filters.
Note 3 - You may wish to enable Spam Reporting (Settings >> General) as any spam reported will update the bayesian filters. Spam Reporting appends a spam reporting link to all non-whitelisted incoming email.
Note 4 - You may wish to add legitimate email addresses to the whitelist to avoid the bayesian filter entirely. That way the bayesian filter will only need to be trained on outlying legitimate emails.
|
Rate this Topic:
|
|||
|
