Document Type

Article

Abstract

Due to the popularity of Internet, e-mail use is the major activity when surfing Internet. However, in recent years, spam has become a major problem that is bothering the use of the e-mail. Many anti-spam filtering techniques have been implemented so far, such as RIPPER rule learning algorithm, Naïve Bayesian classifier, Support Vector Machine, Centroid Based, Decision trees or Memory-base filter. Most existed anti-spamming techniques filter junk emails out according to e-mail subjects and body messages. Nevertheless, subjects and e-mail contents are not the only cues for spamming judgment. In this paper, we present a new idea of filtering junk e-mail by utilizing the header session messages. In message head session, besides sender's mail address, receiver's mail address and time etc, users are not interested in other information. This paper conducted two content analyses. The first content analysis adopted 10,024 Junk e-mails collected by Spam Archive (http://spamarchive.org) in a two-months period. The second content analysis adopted 3,482 emails contributed by three volunteers for a one week period. According to content analysis results, this result shows that at most 92.5% of junk e-mails would be filtered out using message-ID, mail user agent, sender and receiver addresses in the header session as cues. In addition, the idea this study proposed may induce zero over block errors rate. This characteristic of zero over block errors rate is an important advantage for the antispamming approach this study proposed. This proposed idea of using header session messages to filter-out junk e-mails may coexist with other anti-spamming approaches. Therefore, no conflict would be found between the proposed idea and existing anti-spamming approaches.

Download

Share

COinS