1. ProMail: Using Progressive Email Social Network for Spam Detection.
- Author
-
Carbonell, Jaime G., Siekmann, Jörg, Zhi-Hua Zhou, Hang Li, Qiang Yang, Chi-Yao Tseng, Jen-Wei Huang, and Ming-Syan Chen
- Abstract
The spam problem continues growing drastically. Owing to the ever-changing tricks of spammers, the filtering technique with continual update is imperative nowadays. In this paper, a server-oriented spam detection system ProMail, which investigates human email social network, is presented. According to recent email interaction and reputation of users, arriving emails can be classified as spam or non-spam(ham). To capture the dynamic email communication, the progressive update scheme is introduced to include latest arriving emails by the feedback mechanism and delete obsolete ones. This not only effectively limits the memory space, but also keeps the most up-to-date information. For better efficiency, it is not required to sort the scores of each email user and acquire the exact ones. Instead, the reputation procedure, SpGrade, is proposed to accelerate the progressive rating process. In addition, ProMail is able to deal with huge amounts of emails without delaying the delivery time and possesses higher attack resilience against spammers. The real dataset of 1,500,000 emails is used to evaluate the performance of ProMail, and the experimental results show that ProMail is more accurate and efficient. [ABSTRACT FROM AUTHOR]
- Published
- 2007
- Full Text
- View/download PDF