Second-generation Artificial Immune System In The Spam Filter Applications

With the growing use of E-mail, it is urgent to resolve the severe problem of Spam. The traditional anti-spam technology has low classification accuracy rate and has not dynamic adaptive. So, the researchers put the existing immune-inspired techniques to anti-spam field and achieved certain results. But the true positive rate and the true negative rate can not balance well.At the field of artificial immune, algorithms based on danger theory has been made more prominent results in the 2002. Because of intrusion detection and spam filtering has more similarity, this thesis applied algorithms based on danger theory to spam filtering and improved the algorithms. In 2007, the rise of second-generation artificial immune system provides an important idea to improve the accuracy of artificial immune algorithm.This thesis based on the scientific and technological project (serial number of the project: 03GG-006-021). From the limitations of first-generation artificial immune systems and the disadvantage of existing spam filtering technology, dendritic cell algorithm as the main line, dangerous theory and second-generation artificial immune system as the theoretical foundation, this thesis put forward a second generation artificial immune system based algorithm implementation and successfully applied to the field of spam filtering. The improvements of existing algorithm and the proposed algorithm can be used for the classification of other necessary areas, such as text mining, intrusion detection, pattern recognition, etc. Main research contents:1. By generating dangerous signal, applied the latest research results of danger theory dendritic cell algorithm to spam filtering.2. By improving the way of antigen generating,the order of antigen-presenting and antigen-presenting, allows a higher accuracy rate of dendritic cell algorithm.3. Put dendritic cell algorithm and the clone selection algorithm based on the traditional artificial immune system together to form a complete second-generation artificial immune system-based dendritic cell algorithm with training (T-DCA), to further improve spam filter accuracy.