Research And Implementation Of The Anti-Spam Model Based On Transparent Proxy

With the development of Internet applications, E-mail has become an indispensable communication tool in people's work and lives. At the same time, the emergence of spam which destroyed the harmony of network civilization, wasted Internet resources and Internet user's time, has aroused great concern. Anti-spam technology has become a research focus of network security field. Already existing anti-spam products can not effectively curb the spreading of spam. Research and development of effective anti-spam model has become imperative. This thesis analyzed the advantages and disadvantages of existing anti-spam technology, referred to the development trend of today's anti-spam technology, proposed an anti-spam model which based on transparent proxy.Bayesian classification algorithm which based on Bayesian theory was widely used in the field of text classification. In this thesis, I studied Bayesian classification algorithm and its' application in spam filtering. For the localization that Bayesian algorithm may misjudge normal mail as spam that brought huge losses to the user, this thesis referred to the minimum risk decision theory, improved the Bayesian mail filtering algorithm which can effectively reduce the probability of misjudgment spam and improve the accuracy of spam. Transparent proxy server can be deployed between client and server, caching the data packets which through the proxy for detection. This article studied the principle of transparent proxy works seriously, on this basis, the anti-spam model which based on transparent proxy was proposed. E-mail detection and mail transparent proxy modules were made a presentation focusing in this thesis. Mail detection module used the improved algorithm. It can be divided into training and classification sub-modules. From the protocol point of view, mail proxy module can be divided into SMTP proxy and POP3 proxy sub-modules. Mail proxy implements that caching e-mail message content for detection and executing specified action on mail that detected. Finally, Experiments and tests of the model were introduced with the summary.The result of research in this thesis is implementing an anti-spam model which based on transparent proxy. The model can detect spam accurately, executive specific action on spam according to the configuration of manager. Experiments and tested proved that the improved Bayesian classification algorithm can improve the accuracy of mail classification, Mail proxy achieved expected requirements that caching e-mail content for testing and performing specified action on detected mail which is spam.