The Method Study Of Phishing Defense Based On Sensitive Characters

With the development of network, the network is becoming the platform for people to work, to learn, to shop and also other activities of financial transactions.etc, which has brought great convenience to the people’s working and living. However, with facilities of network, the network crimes occur frequently. The overflow of phishing attacks, brings about a great deal of distress to people’s normal life, also results in adverse effects of the trust between people and people, and the trust between people and legal institutions. It especially impacts the industry of e-commerce and e-finance, blocks the development of electronic commerce. Therefore, there is an urgent need for us to study the detection of phishing attacks.Phishing attack is a kind of social engineering, which uses people’s psychological weakness, to fool them by meet their desires or wills. In the view of behavior psychology, this paper analyzes the purpose of phishing designed by attacker, and the psychological action when people operate on the website, sets up phishing model based on the theory of motivation behavior; according to habituation of learning and sensitization by stimulation, analyzes habituation of people operating on website, and sensitization occurring when stimulation of phishing attack, proposes behavior model of phishing. Based on this model, this paper proposes some improvements of phishing detection in the chapter fifth, and also some user-self-protection phishing comments are given.The present tools for phishing detection have two methods mainly:blacklist and heuristic feature detection. However, blacklist is always lagged, without updating malicious site in time; the features of traditional feature detection are targeted, so it always prone to give false justice. In the fourth chapter, this paper proposes a method of phishing detection based on sensitive characters, extracts the features of URL and other identity, then classifies the website with a linear classifier. The results show that sensitive features extracted from URL and webpage is useful for distinguishing phishing artificially, and the detection method gives a higher true positive and false positive.