Help: Configuring Emmesmail to avoid spam

The parameters to tailor how Emmesmail treats spam mail are found in the advanced section of the POP3 configuration menu. Simply go to the main Configuration menu (log in under your user name and choose configure from the next menu). Then choose "POP3" and then "Advanced".

The first thing to do is to decide how agressively to treat spam. Choosing a "Spam action" of "0" results in no spam filtering. We recommend initially setting the "Spam action" parameter to "1", which causes spam mail to be deleted from the server and delivered only to user spam. It is also possible to set the "Spam action" parameter to "2" (discussed below).

The next step in configuring Emmesmail to reject spam is to select the anti-spam features you wish to use. The two choices are "sender filtering" and "Bayesian filtering". We recommend always using sender filtering. Using sender filtering causes the sender name to be compared to names in your local whitelist and blacklist. The blacklist and whitelist, respectively, are lists of senders you previously determined to be spammers, or non-spammers. The use of the blacklist helps prevent you from receiving two spam emails from the same sender and the use of the whitelist reduces the possibility of the occasional false positive result. It shouldn't be necessary, but if either list gets corrupted, the whitelist may be edited by going to the main Configuration menu and clicking on "White", and the blacklist, by clicking on "Black".

Bayesian filters work by putting the spam emails you have received into one corpus (database), and the normal emails you have received into another corpus. They then create from the corpi a hash table which gives for each word, the probability that the word is appearing in a spam email. When a new email arrives, each word of the new email is looked up in the hash table and the probability the new email is spam is calculated. Until there are enough emails in your spam and non-spam corpi to create a meaningful hash table, Emmesmail uses Emmes Technologies' hash table, which works excellently for us in rejecting what we consider spam, but may not be so well tailored for your preferences and the kinds of spam emails you receive.

We have no doubt that once you have designated a sufficent number of spam and non-spam emails so they get saved in your spam and non-spam corpi (see the information on training), Bayesian filtering will work for you with excellent results. The question arises as to how soon to select the Bayesian filter option. We have the following suggestion.

If you are currently receiving large amounts of spam, we recommend selecting the Bayesian filter option almost immediately, and checking the mailbox of user "spam" frequently for false positives.

If you are not currently receiving many spam emails, either because you have been very careful, have a new email address, or your ISP does spam filtering, we recommend selecting the Bayesian filtering option only once the number of spam emails you are receiving becomes bothersome.

The filter is "trained" each time you designate an incoming email as valid (by saving it), or as spam. This occurs regardless of whether Bayesian filtering is on or not.

Once the spam filters have been trained, you may leave the "Spam action" parameter at "1", or set it to "2". When the "Spam action" parameter is "2", whenever a spam email is rejected, an email is sent back to the sender of the email, the spammer, with the error message, "Emmesmail does not accept spam". The error message may be customized by editing the file <install-dir>\config\failmess.txt. If this file does not exist (by default it doesn't), you may create it by copying the file <install-dir>\data\failmess.txt to the config directory and then edit <install-dir>\config\failmess.txt. It is controversal whether sending emails back to offending spammers decreases, or actually increases, the amount of spam, but we include this option should you wish to use it.

If you receive spam from iytdf@wesendspam.com, hgfvn@wesendspam.com, and sfgnq@wesendspam.com, it is quite clear that you want to filter out all mail from the domain wesendspam.com. The way to do this is to edit your blacklist and make sure it has the entry "*@wesendspam.com". You may, at the same time eliminate the entries for iytdf@wesendspam.com, hgfvn@wesendspam.com, and sfgnq@wesendspam.com, which presumably were added when the previous spam emails were detected, and will no longer be necessary.

Read about handling spam and training the Bayesian filter


Back to main help menu