Search

This paper applies Tesseract-OCR, optical character recognition software, to image spam mail filters. Tesseract-OCR can be specific to a certain language and this paper makes Tesseract-OCR specific to spam words. This specialization decreases times and CPU power that it takes to check whether images of mails include spam words. This paper examines the ability of the spam mail filter of Tesseract-OCR by experiment.

Today e-mail is a indispensable communication tool because of wide spread of Internet technology and many spam mails are delivered in the Internet. In order to avoid spam mails, spam filter is used in many mail servers or PC. It checks the content, sending address or so on of a mail and decides whether the mail is a spam mail. After spam filter works well to remove spam mails, spam senders improve the methods of sending spam. Spam filter technology and spam sending method make a vicious circle. Ending this vicious circle needs spam mail characteristics including the change according to spam filter improvement. This paper analyzes spam mails which are saved in the long term in order to obtain spam mail characteristics and reports the result of this analysis.