| 摘要 |
In recent years, the use of spam zombies has become a preferred method of sending spam. In fact, it is estimated that over 90% of all spam comes from spam zombies. Although existing spam zombie detection mechanisms such as the Spamhaus XBL blacklist exist, these techniques are limited in that they cannot block spam from newly created spam zombies. The present invention relates to a system and method for detecting malicious e-mails from spam zombies, the system comprising a processor operable to process a server identification value of a sending source by separating the value into one or more domain level terms to allow each unique term to be tokenized with an index value and to apply the one or more tokenized values as a learning feature in a learning algorithm trained to identify spam zombies.
|
