How is SPAM identified?
Our Spam filter uses several different methods to identify SPAM. External SPAM databases Vipuls Razor and DCC are used to build a checksum of all incoming email and uses this checksum to verify the email against a database of known SPAMs. If someone else participating in Vipuls Razor or DCC projects has already received this particular SPAM, then the presently checked mail will be marked as SPAM as well.
Additionally, emails are weighted. Features typical of SPAM messages will increase the SPAM-rating, while other factors like those you usually only see in legitimate mails decrease the rating. A "whitelist" is also kept, so if a person has sent three (configurable) legitimate emails to you so far, then further emails from this person are less likely to be identified as SPAM.
Bayesian filtering is also used, which effectively makes the filter self learning by evaluating previous legitimate emails and SPAM messages.