HOME  >  Research  >  ナナサメ・癸シ・・ユ・」・・ソ・・・ー  >  Summary of the E-mail filtering

E-mail filtering

The Internet or a cell-phone become imminent, and, with development of the recent communication technology, the E-mail comes to be essential to life. A supplier sending an advertisement or a phishing email using the E-mail for the unspecified number of person called the spam mail appears, and it is with a problem while many people use it. The ratio of spam mail as of March, 2013 accounts for more than 60% of all emails. (figure 1)


figure 1。。ニヒワ、ヒ、ェ、ア、シソョ・癸シ・、ホ・ケ・ム・爭癸シ・。ヲタオオャ・癸シ・、ホウ荵

Because the original spam mail was the simplicity that included a specific e-mail address and server name, coping was simple. However, I displayed the e-mail address of the third party and an unstable e-mail address and disappeared recently using a specific e-mail address. 

There is a Bayesian filter ring in the email filter used widely now. The word calculates the frequency included in the spam mail using Bayesian theorem from the word included in the email and I calculate email spam probability for judgments based on the frequency and, based on a spam mail county and the regular email county that I classified beforehand, judge it to be it. 

However, the contents of the recent unwanted e-mail change diversely, and it is necessary to update the classification of the word in a fixed period of time. 

Therefore I suggest the technique that I reduce the trouble of the person and classify words in by using machine learning system BONSAI in this laboratory. 

 

クスコ゚ソハケヤテ讀ホクヲオ

。ヲナナサメ・癸シ・・ユ・」・・ソ・・・ー、ヒ、ェ、ア、オ。ウ」ウリスャ・キ・ケ・ニ・滷ONSAI、ホウリスャホ譱、ネタコナル、ホエリキクタュ

 

ウリイネッノスマタハクセハ、マ、ウ、チ、


MENU