Spam Filtering Based On Artificial Immune System And Bayes Network

Today, E-mail has become one of the fastest and most economical ways of communication available. Spam is an ongoing problem on the Internet, in despite of E-mail is being the convenient communication means. People receive more spam E-mail than the normal E-mail. So spam filtering is a subject with great realistic meaning. Artificial immune system (AIS) strong self / nonself discrimination ability (Immunity Discerning) is a good solution to spam filtering. The sensitive immunity feedback mechanism offered the solution for discerning the changeable spam E-mail. How to apply these metaphor mechanisms of the Artificial Immune System to the spam filtering , which is the subject matter that this paper will solve. Firstly, the paper analyzes the principle of E-mails ,the harm of spam E-mails and the Current situation of the spam filtering, it introduces the corpus for spam filtering and the appraisement system of spam filtering. On the basis of analysing the existing spam filtering technology, the paper proposes a new spam filtering means based on Artificial Immune System and Bayes network. Secondly, the paper analyses the basic principle AIS and Bayes network. It not only analyses the metaphor mechanisms of AIS, but also describe the framework of immune algorithm and common immune algorithm. On the other hand the paper introduces the topological structure of Bayes network, assumption of the independence condition, the basic Bayes formula. Finally, a spam filter algorithm based on AIS and Bayes network has been proposed. The paper analyses the key problems of the algorithm, and provides solutions of these key problems. A spam filter model based on AIS and Bayes network has been devised,and an artificial device software for spam filtering based on this model has been developed. The artificial device software for spam filtering behaves well in the PU1 bare corpus.