|
With the rapid popularization of Internet , e-mail has become one of the primary means of modern communication . However, the presence of spam is threatening the safe use of the e-mail . Spam as commercial advertising, malicious programs and unhealthy content carriers , not only consume network bandwidth , causing adverse effects on the user a waste of time, money and emotional at the same time , serious interference with people's normal life . Therefore, the anti - spam action without delay , its technology has been steadily upgrading . Currently, there are anti-spam technology , based on the the SMTP layer of anti-spam technology and filtering based on message content and technology R \u0026 D based on the IP layer . Inspired by data mining text mining , dedicated to the research based on the content of the message text spam filtering technology . The first to use the forward maximum matching word segmentation mail sample body text , get mail feature items . Then, the use of mutual information to reduce the vector dimension of building a mail sample library . The principle of the introduction of the more popular Bayes , KNN , vector space , Naive Bayes classification , a comparative analysis of these algorithms . Traditional KNN algorithm slow search speed , sample storage capacity dependent defects , mixed classification model based on KNN algorithm . The algorithm first use of the classifier to process the mail classification results , according to the classification results the same and different decisions again KNN calculation , thus avoiding the limitations of the single classifier , can play the advantages of each classifier . The experimental results show that the method for Chinese e - mail filtering system is feasible , to good effect , is a valuable attempt a mail filtering . Finally , this paper presents the multimedia spam filter , the phone spam SMS filtering the direction of scientific research needs to be further open up .
|