Research of the Anti-phishing Technology Based on E-mail Extraction and Analysis

In this paper, an anti-phishing technique based on e-mail extraction and analysis is proposed. The technique approached with phishing email, the channel phishing attack transmits, distinguish phishing emails and extract the suspicious URL from the e-mail for further analysis. Upon arrival, a protected list is built according to those third parties which are the most vulnerable to phishers in order to filter those confusing advertising spams in China and a neural network based model is proposed in order to detecting phishing messages from an e-mail stream. In this anti-phishing technique, email stream captured by our honey pot subsystem from the Internet is parsed into a MIME email firstly, various feathers are extracted from the email and outputted into feather vectors. The feature vectors will be self-organized by ART2 neural network one by one and classified into corresponding categories. Link URLs in the suspected emails´ messages will be extracted for further detection in phishing site subsystem. Experiment using collected emails shows a good performance aiming at detecting phishing emails in China while foreign method performs badly in distinguishing phishing emails from spam.