| Nowadays the increasingly frequent used in the network, e-mail exchanges gradually become the main tool. But, more and more spam took a lot of network bandwidth, affecting the normal mails transmission, and to people's daily lives brought many inconvenience. With people keeping in the Anti-Spam, the extension of Spam is to expand to the mails which include viruses, worms and Trojan horses. And spammers'technology, the spread of the worm virus technology, all forms of intrusion in line with each other, all that makes spam more and more hidden, pluralism showing and a integration of the situation. As for the variation of the characteristics of spam, traditional anti-spam approach is insufficient, the study and design to adapt to the ever-changing spam technology a filter algorithm appears to be particularly important.With "adaptive, self-organization, self-learning" and other characteristics of the evolutionary thinking of the algorithm we study characteristics of the ever-changing spam provide an important direction. This thesis includes the work I have completed: the study of the current anti-spam technology, an evolutionary thinking of the Bayesian algorithms and artificial immune algorithm on the basis of the design and realization of the anti-spam algorithm testing platform, as well as artificial immune anti-spam algorithms and Bayesian algorithm, the use of CCERT corpus, and the test platform on the two algorithms were tested. Algorithm test consists of a separate test and the horizontal test algorithm. Longitudinal test major against artificial immune algorithm that its experimental results show that artificial immune algorithm in the number of lymphocytes more, it does not mean spam recognition rate higher, and the algorithm, tend to 0.34 threshold, spam recognition rate is the highest. Horizontal test results showed that the artificial immune algorithm than Bayesian algorithm, more adaptive, stability and the higher correct recognition rate. |
PDF Full Text Request