近來垃圾信(SPAM)數量近來成長的相當快,目前常用的方式是使用黑名單(blacklist)或使用關鍵字(content filter)去過濾垃圾信件,如使用貝氏定理去分析信件內容產生過濾關鍵字的貝式過濾。但是對於發送垃圾信的有心人士可以對信件的收件者發送位址做偽造或加入額外的內容來躲避掉過濾,並且容易對於正常的信件造成誤判,讓使用者花費額外的時間到垃圾信匣中找回正常的信件。相對於傳統的過濾方式,我們設計了一個系統,並使用郵件認證(E-mail authentication)的方式,去大量減少使用者收到垃圾信的數量,並且減少誤判發生的機率。每個使用者對於特定垃圾信的定義有可能不同,例如像廣告信件,因此讓使用者自行決定寄信者是否合法的方式,這是目前常用的過濾垃圾信件方法中沒有的。系統遵循標準的SMTP協定,所以不會影響到我們和網路中沒有安裝該系統的郵件伺服器的溝通,且不需對於使用者的MUA(Mail User Agent)做任何改變,希望達到對於使用者使用習慣影響降到最低。
Recently, the number of spam has grown up quickly. Blacklist and Content Filtering are the most commonly used anti-spam systems. For instance, Bayesian filtering-it is a process of using Learning Bayesian Classifier to classify keywords from mail content to distinguish spam. However, most spammers would forge sender addresses or add extra contents to evade content filtering. Therefore, when Content filtering system filters e-mails, it might incorrectly classify legitimate mails as spam, which would take recipients more time to search legitimate mails from trash box. Different to traditional anti-spam systems, we designed a system, which use E-mail authentication to effectively reduce the number of spam and false positive rate. Our system enables users to judge whether the sender is a legitimate user or not. Our system follows standard SMTP protocol; therefore, mail servers without installing our program can still communicate with our system. It is unnecessary to install extra programs in User Mail agent. We endeavor to lessen the influence on users’ using habits to the minimum.