Spam Control
This online spam control service has been developed to assist WPI e-mail users to filter SPAM by taking full advantage of the new spam marking feature now available on our e-mail server. Use this web page to define settings for controlling SPAM that is normally delivered to your WPI e-mail account.
This service can install a number of procedures for handling spam:
- It can white list addresses so that they are forwarded to your inbox, despite blacklist or spam probability.
- It can black list messages so that they are not forwarded to your inbox.
- It can block messages which exceed your tolerance threshold of spam probability.
- It can apply an indicator of spam probability to the subject header of the mail, so that the spam rating is visible.
- Set a vacation message. Note that if you are only setting a vacation message and not using any spam control features, please use this simpler vacation setting procedure.
A black list is made for messages you do not want to see. A white list is checked before the black list, for special messages you do not want blocked. e.g. you might black list aol.com, so you would never see a message from there, but you might white list mom@aol.com so that you can get your mother's email.
Filtering with your own programming
If you want to deal with the filtering on your own, you can choose your own filter software and program it to look for the spam probability marking.
The PureMessage spam marking software inserts a header with its estimate of the probability that a message might be spam. You will not find this header in all messages; only messages which begin to raise a suspicion are marked. The header is X-Perlmx-Spam and the first entry in the header is a Gauge of the spam probability. This Gauge field should be easy to handle with any filter program. The Gauge value is in a simplified roman numeral form, with leading X's for tens and trailing I's for ones. e.g. a Probability of 77% is represented by the pattern Gauge=XXXXXXXIIIIIII. This roman numeral form avoids V and L, which would complicate pattern matching.
It is easy to differentiate spam at any 10% amount, but probably not worth the effort of differentiating at an "odd" amount. e.g. a probability over 73% could be differentiated by the pattern Gauge=XXXXXXXX (any probability 80 or over) or the pattern Gauge=XXXXXXXIII (any probability 73 through 79). i.e. the application of two rules would be needed to specify the fine-grained 73% probability.
If your filter software can deal with a decimal number, the next field in the X-Perlmx-Spam header is a numeric probability display, e.g. Probability=77%.
Both the Gauge and Probability entries are followed by a comma. You should observe some X-Perlmx-Spam headers, to be sure you understand the form, before you install your rules.
Helpdesk has developed descriptions for setting local filtering on a few platforms:
aej@wpi.edu Last modified: Monday, 30-Mar-2009 20:35:34 UTC