Research On P2P-Based Spam Detection System

Mass of spam delivering on the Internet day and night costs mail service providers and organizations lots of money, time and energy for the lost of bandwidth and the spreading of virus. Further more, the usage of e-mail for delivering erotology and reactionism has had fiendish impact on our society. Thus, research on anti-spam technology is of great importance.Current spam detection technique, considering its applied location, can be classified into two types: server-side and client-side, both of which are centralized, and are difficult to attain sufficient spam information from other mail servers or mail clients. Considering the detection process, current spam recognition technique confirms to such rules: captures a mail, generates a rules, conveys the rule, and filters by mail content. The main drawback of this process is spam has to been delivered to mail server before filtered, and thus network bandwidth and resource has been consumed. A spam detection technique which shares spam information among mail servers and stops spam before it arrives mail server is badly needed.P2P (Peer to Peer) is an application pattern in which resources and services are shared between nodes. Each node acts as sever and client as well. In the last few years, P2P is greatly developed due to its equality, open and freedom. Aiming at making up the shortages of current spam recognition technique, a P2P-based collaborative spam detection system is designed in this paper. It shares spam detection information through P2P network. Also, a spam SMTP (Simple Mail Transfer Protocol) behavior recognition subsystem is designed based on researches on the differences between spam and legal mail during the SMTP process. The subsystem is able to hold up apparent spam before it arrives at the mail server. A mail digest algorithm called Nilsimsa is introduced and utilized in this paper. An improvement aimed at raise Chinese spam detection efficient and accuracy is made and tested.