One of the features that Help Desk has long supported is spam filtering, a constant concern when one publishes the email addresses used by your help desk system to the open web. In version 3, we reworked the algorithm to one where the spam filter could learn from its mistakes and tailor itself to fight the kind of spam that your site, individually, tends to receive. We'll explain in this article how the spam filter works and how you can best use it to fight spam.
The core interface to our spam filter in version 2 and 3 of our Help Desk software is essentially the same. When you see a ticket that appears to be spam, you can check it on the ticket list and mark that ticket as spam. Alternatively, you can click the "Mark as spam" button when viewing the contents of an individual ticket. Marking a ticket as spam will cause our software to parse the ticket, looking for interesting words and retaining information on the number of times each word has previously been marked as spam. It'll use that information in the future to help automatically detect if a message is spam or not.
Our spam filter starts from scratch, with no knowledge of the kind of email that you get. Once you start marking tickets as spam, it'll do its best to learn, but what will help it the most is to mark as many tickets as spam that you can. Occasionally (especially once you first get going) you will have a false positive, which is a ticket that was automatically marked as spam by the system but which isn't really. When you see that, mark that ticket as not spam (again, either through the ticket list or viewing the ticket individually); doing so will again help our system learn the differences between spam and non-spam for your system. Responding to a ticket will also cause your system to learn that ticket as non-spam, so that you don't need to wait for false positives for it to begin learning.
The more you mark tickets as spam, the better your system will get at finding the right spam tickets you receive and letting other tickets through.
