Font Size: a A A

The Research Of Spam Filtering Technology Based On Artificial Immunity System

Posted on:2009-02-03Degree:MasterType:Thesis
Country:ChinaCandidate:H ZhuFull Text:PDF
GTID:2178360272977141Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Because the problem of spam has became seriously, the paper introduces a spam filtering method based on AIS(Artificial Immune System), The idea of immunology was adopted to build an anti-spam E-mail system, especially to the spam that are dynamic and variant. Compare with the common filtering method, the immune technology has the features of auto-learning, auto-adaption and robustness.Firstly, the paper studies some kinds of present spam filtering technologies nowadays, by means of comparison with these methods, the paper discusses the dynamic and auto-adaption characteristics of anti-spam filtering method that is based on AIS; secondly, the paper studies the principle of AIS, analyzes and discusses the common features of AIS and spam filtering system, after that, an anti-spam system model based on AIS, which is called ECSBAI(E-mail Classify System Based on Artificial Immune)is presented, and then, the algorithms of key modules are given. Trough the method of experiment, including comparison with Bayes algorithm, the paper analyzes the effect of selection of key parameters on the filtering result and the feature of AIS that prevent variant spam. Thirdly, to the question that inefficiency of detector production, and the shortage of detection precision, the paper designs an algorithm to produce detector which is based on weight of gene, and then improves the algorithm of spam filtering, and the result has illustrated that kind of efficiency. Finally, the paper discusses the technology of Trojan detection based on AIS, and to the feature of that Trojan varies rapidly, a detection model based on AIS is proposed, and the design of key modules is also given,and then, though emulation experiment, detect effect of the model is validated.The main innovation of this paper is the improved algorithm of detector production and spam filtering, and the experiment has illustrated that, compared with the former method, the detection efficiency has advanced to a certain degree, especially to the novel spam.
Keywords/Search Tags:spam, artificial immune system, negative selection algorithm, clone selection algorithm, Trojan detection
PDF Full Text Request
Related items