Information Systems & Services (ISS) - spamassassin

Information Systems & Services (ISS)

spamassassin

Spamassassin is a program which analyses mail messages and attempts to identify spam. We use this program in DCU to tag your spam and warn you about suspected spam content.

Tagging Your Spam

Spamassassin warns you about mail which may be spam. Ordinarily you might receive a spam mail with the subject line e.g.

  • Guaranteed Weight Loss

Now the subject line of the same message would be changed to this

  • { *** SPAM *** } Guaranteed Weight Loss

This allows you to filter your spam so that it does not clog up your inbox.

Mistakes

This computer program (spamassassin) is given the task of deciding if a mail is spam or not. This is quite an easy job for a human but not so for a computer. Therefore the system is not perfect and will give rise to a small number of mistakes.

False Negatives
This is a mistake by the program where it failed to identify a spam mail as spam. The mail will not have { *** SPAM *** } in the subject line. Therefore your filter will not have worked either. This type of message will appear in your inbox. This should be deleted without opening/reading. Spamassassin is over 90% accurate in identifying spam. So you will receive some messages in this category. We will try to keep them to a minimum.

False Positives
In this case a message which is not spam has been mistaken as spam. It will be a genuine message intended for you but will mistakenly have the subject line changed to start with { *** SPAM *** }. Therefore if you have your
filter setup this message will not appear in your inbox. It will appear in your spam folder. This type of mistake is more serious than false negatives because it is possible to lose genuine mail. It is essential to follow the guidelines for setting up your filter. Do not set up a filter that will automatically delete spam. You will need to check your 'spam folder' periodically for genuine mail. This type of mistake will not happen very often. It may happen once or twice in every one thousand messages you receive.