Information Systems & Services - Fighting Spam at DCU

Information Systems & Services

Spam System FAQ

Why do I receive some spam into my inbox (false negatives) ?

The copperfasten system uses complicated software (spamassasin) to classify millions of messages every month into 'clean' and 'spam' categories.

Unfortunately no spam system is perfect and it is very difficult to classify so many messages with 100% certainty. The software works by anyalysing the message and assigning a spam score. There is a cutoff score (currently 5) and anything above this score is considered spam and put in the quarantine. Anything below this score is considered clean and delivered to you.

There are some statistics here on the performance of this system. This report from November 2008 shows that of over 35 Million messages scanned that month only 3.2% was considered clean mail and delivered to user inboxes. Over 96% of all mail was considered spam.

Of the 3.2% 'clean mail' there will be a small percentage which is infact spam (false negative). There is always a trade off between false negatives and false positives. Lowering the spam cutoff score (5) reduces the number of false negatives (catches more spam) but increases the number of false positives (quarantines more clean mail).

While it is the goal of all spam filters to have 0 false negatives and 0 false positives, this is very difficult to achieve.