It has been upgraded from 5.2.X to 6.1.X. This new version has a little bit of a new look but everything functions the same. Most importantly, however, it has a new Junk Mail feature that many users have been asking about.

Take a moment to get familiar with this new feature by reading a few tips and strategies to effectively using it on a day-to-day basis.

The Junk Mail Option Explained
The Junk Mail option of the new version of Eudora implements a Bayesian filter. Unlike traditional filters that look for specific keywords in the subject or body of email messages, the Bayesian filter uses the entire context of a message to identify a message as spam. It does this by comparing the words and phrases in the message to probabilities that these words/phrases are found or NOT found in spam.

Bayesian filters have two key benefits over traditional, content-based filters. First, they are adaptive; as spam evolves, so does the filter. You do not need to continously add or revise filters because the filter changes itself. Second, they get more accurate the longer you use them.

Unfortunately, you do need to spend a little time 'training' the filter in order for it to be effective. 'Training' a Bayesian filter is done simply by manually telling the filter which messages are junk/spam and which messages are not. The filter will become more accurate as more messages are properly categorized. Unlike some mail clients that require a certain number of messages to be categorized before it will automatically junk messages, Eudora will start junking messages immediately.

Messages that a Bayesian filter tags as spam are, then, transferred to a Junk mailbox for review. As with any spam filtering method it is always good to periodically review the Junk mailbox for false positives.

Sit!!! - Training Eudora
The act of 'training' a Bayesian filter is the process of manually tagging spam as junk and legitimate email as not junk. It is also something that you'll never quit doing if you want to effectively use the filter.

In Eudora, the process of tagging a message as junk is relatively simple. Right-click the message in the preview pane or in the mailbox table of contents. From the context menu, select Junk (diagram two, #1). The email message will instantly receive a junk score of 100 and be transferred to the Junk mailbox.

To tag something as Not Junk, simply repeat the process and select Not Junk from the context menu (diagram three, #3).

TIP #1: The process can be sped up by selecting multiple messages in the table-of-contents by control- or shift-clicking to highlight multiple messages.

TIP #2: A keyboard shortcut for junking a message is CTRL-J when that message is selected.

False Whats?!?!
A Bayesian filter will do its best to properly junk messages... but undoubtedly during the first few days it'll make mistakes. Messages that are spam will not be junked and messages that definitely are not spam will be. These are called false negatives and false positives respectively. The more time you spend training the filter the less mistakes it will make.

When the filter makes a mistake, what do you do?

For those false negatives, spam that is not junked (still in your In mailbox), train the filter by manually tagging the message as junk. Right-click the message and choose Junk from the context menu. You can also use the keyboard shortcut of CTRL-J once the message is selected/displayed.

For those false positives, legitimate email that ends up in the Junk mailbox, train the filter by manually tagging the message as not junk. You'll want to periodically check the Junk mailbox for false positives. Marking a message as Not Junk in the Junk mailbox should move the message back to the In mailbox.

By correcting false positives and false negatives you are helping train the Bayesian filter in order to improve its accuracy. The more you train, the less false positives/negatives you should have.

TIP #3: A quick method to spot false positives in the Junk mailbox is to look for comparitively low numbers in the 'Spam Score' column (diagram three, #2).

Diagram Three

Other Available Options
There are some additional settings that can be changed using the 'Junk Mail' tab of the 'Options' control panel (diagram one). You can find the 'Options' control panel by choosing 'Options' from the 'Tools' menu. A few of the more important settings include:

Junk Threshold
The Bayesian filter has a scoring system with higher numbers corresponding to a higher likelihood that the message is spam. Decreasing the threshold will filter out more spam (but increase false positives) and increasing the threshold will decrease false positives (but filter out less spam).

Mail isn't junk if the sender is in an address book
Check this box to indicate that mail coming from people in your address book(s) should never be junked.

Put not junked senders in address book
Check this box to indicate that senders of valid mail found in the Junk mailbox are to be added to your address book when you select Not Junk from the right-click context menu. Doing so will whitelist this email address (future email from this address will never be junked. (This function only works if the previous setting is also checked.)

Automatically place junk in Junk mailbox
Check this box if you want Eudora to filter what it considers junk mail to the Junk mailbox when mail is downloaded. If this is not checked, junk mail appears in your In mailbox and is scored. To see the junk mail score, go to the Mailboxes option in the Tools menu and check the Junk box (located under the Show Mailbox Columns section).

There are quite a few other settings that you can experiment with to modify the behaviour of the Junk Mail feature. You can find these on the Junk Mail and Junk Mail Extras option control tabs. Additionally, this version of Eudora has some new filter actions to assist in training the Bayesian filter. If you created filters to manually transfer spam messages to the trash, modify these filters to junk the message instead.