Spam Filtering System Based on Uncertain Learning

Anti-spam system is asked urgently in recent years. Compared with traditional anti-spam system, a new spam filtering system is proposed in this paper which is based on uncertain learning approaches. These uncertain learning approaches are well integrated through a commission collaboration mechanism. The new system could handle dual-way spam filtering, namely both out-going and in-coming spam filtering. Six-month performance test on a real email server proved that the new system has very low FN and FP ratio.