The technology of spam filters has recently received considerable attention as a powerful approach to Internet security management. The traditional spam filters almost adopted static measure, and filters need be updated and maintained frequently, so they can not adapt to dynamic spam. In order to get over the limitation of the traditional means, an immunity-based spam classification was proposed in this paper. A brief review about traditional technology of spam filter is given first, particularly, the Bayes probability model. Then the NASC is described in detail by introducing self, non-self, detector and detect algorithm. Finally, a simulation experiment was performed, and the important parameters of the model were analyzed. The experimental result shows that the new model increases the recall of filter greatly in condition that precision also increasing, and demonstrate that the model has the features of self-learning and self-adaptation.
展开▼
