Enabling the local Bayes training can help with filtering, but it does require active interaction to train the system. With local Bayes training, you can designate an email that you have received in your Pending quarantine as spam or non-spam, which will help tailor the filter algorithms to detect the type of spam that you personally receive. The following steps are used to designate an email as spam or non-spam:
1) Log into your Roaring Penguin account at https://antispam.roaringpenguin.com/canit/index.php
2) Click on the Date link of an email you would like to designate.
3) On the Bayes Training line, under "Incident", click on either the "Train as spam" or "Train as non-spam".
The more emails you designate as spam or non-spam, the more effective the Bayes training will be. However the system does require a fairly sizable amount of training before it will begin to use the local Bayes list. While this method tends to be more accurate as far as spam detection goes for an individual user, it does require a lot more interaction to reach that level of accuracy.
Another tactic that can help reduce the amount of spam you receive with less interaction required, is the use of Custom Rules. When you see batches of spam with common factors (such as references to "Rachel Ray" in the email), you can add custom rules that adjust the spam filter scoring according to those factors. The following knowledge base article will guide you through the creation of custom rules: