A study of spam mail filtering based on transmission behaviors
| Autor: | CHIA-LUN LIU, 劉家倫 |
|---|---|
| Rok vydání: | 2006 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 93 The Internet is getting more and more popular. But the unsolicited electronic mails(also called Spam)in Taiwan become a very important issue which is highly valued by government and the general public. According to the ASRC(Asia SPAM-mail Research Center) and The Internet Institute in Taiwan, the “Survey of Misery Index in Spam-mail” in 2004 showed that Internet users receive about 10 to 50 junk mails each day. They make up a large proportion to 43 percent. Secondly, the proportion of receiving 50 to 100 junk mails each day makes up to 25 percent; the proportion of receiving more than 200 junk mails each day makes up to 6 percent;. This survey found that the average misery index for the people in Taiwan is up to 71.76 percent. It is very hard to estimate for the losses of time and money. Nowadays, the technology of filtering spam can be divided into three general orientations. The first one is the technology of content filter which include Email header and the keywords of content. The second part is to use the standard communication protocol as the judgment of transmission behaviors. The third part uses the domain keys of e-mail box to identify senders’ mail server, but it still several problems remain to be solved. This thesis focus on the attacks of Email spoofed and Open Relay. First, it uses the transmission message of e-mail header as the judgments on the spam-mail, illegal, anonymous, forging behaviors, etc. Then it utilizes the database of spam-mail in the website to and Perl program to intercept and compare each other. Finally, through bagging, support vector machine and algorithm of LMT (Logistic Model Trees ), the experimental result reaches the anticipated effect. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|