SpamExperts combines several technological concepts; a self-learning loop is created which dramatically enhances the filtering results. At the client level, when an email is received, the content is statistically analyzed (based on the Bayesian algorithm). If the content is determined to be significantly similar to spam, the local database storing spam content information is updated. If the content is not determined to be significantly similar to spam, the local database storing ham (no spam) content information is updated. This technology by itself has already proved to result in high filtering rates with low error margins. The main disadvantage, however, is that the software will first need to learn content in order to distinguish between ham and spam. Without content training for each individual user, the software cannot significantly judge any email. In order to overcome this problem, another technology creates from every email a unique identification string, stored in a central database. This concept is similar to a human fingerprint that can uniquely identify an individual. The self-learning algorithm of only a few users will automatically identify new spam emails, and train the central database with this information, thereby sharing it with all other users. This way, the system will automatically anticipate new spam emails collaboratively, and therefore outsmart the spammer without human intervention! Please note that the system uses very sophisticated methods in valuing the incoming reports on spam and ham, and drawing conclusions on these reports. The system does not work as black and white as described above. Many additional techniques ensure the software to function on extreme low error-margins. - Automatic email interception - Compatibility with all POP3&IMAP based email clients - Compatibility with Windows 2k, XP and 2003 - English, Dutch, Spanish - Cyrillic character support
Comments