A Collaborative Anti-spam System Based on First-Second Filtering
Xiushuang Yi · Journal of Xiamen University · 2007
A three-level anti-spam model and a novel antispam function framework based on first-second filtering are introduced in this paper.In the first pass,rule-based filtering method is used to process the mass mails into three classes: ham and spam and suspected spam quickly.In the second pass,content analysis method is used to discriminate the suspected spam further and the adaptive method based on feedback learning is used to update the rules used in the first pass.An anti-spam system has been implemented based on the above model and framework.Experiment results have shown that it is effective and efficient.Its miss rate is lower than 3.7%,false rate is lower than 0.6%,and the performance is better than other anti-spam products.