Identification Of Neural Network-based E-mail Classification

With wide application of email, spam, acting as the carrier of business advertisements, the malicious programs or some sensitive mails, are more and more fiercely threatening the safety of the computer systems and the lives of people. Anti-spam problem has become an international, significant and practical topic now.However, with the fast increasing of spam, the anti-spam techniques have developed slowly. Now a majority of anti-spam techniques lack of aptitude and auto-learning, which can not identify new spam by learning from former spam instances. And a few anti-spam techniques have auto-learning character, such as Bayes algorithm, it only work on the content of email and ignore email's head fields, which is the most shortage of the technique.Suggested from the anti-spam technique of Bayes algorithm, I analyzed the shortage of the traditional anti-spam techniques. On the basis of studies and statistics on a lot of emails, my paper brings forward mail classification and recognition model based on neural network. The model synthetically analyzed the characteristics of spam and legal emails and disperses and characters emails then distill nine characters to express emails on vectors by programming. BP algorithm has aptitude and auto-learning characters, so my paper choose BP neural net algorithm to set up mail classification and recognition model. My paper compared and analyzed some familiar betterment of BP algorithm; at the same time choose the betterment of LM to enhance performance of the model according as rapidity and precision of the model. At last the model is emulated and tested by programming. The result indicated the model can find how to identify the new spam by learning from the field, network, structure and content information of emails. It shows mail classification and...