基于特征空间异化重构的Winnow算法过滤网络钓鱼

Phishing is increasingly becoming a threat to network security and network users suffer from huge economic losses. This paper presents a method to filter phishing using the Winnow algorithm based on feature space alienation reconstruction。It extracts sensitive features of Web pages by analyzing them, and then makes classification (normal web pages and phishing pages) using Winnow algorithm. At the same time, it makes feature space reconstruction for the alienation through amendments to Winnow algorithm parameters by BP neural network, all of which improves the Winnow algorithm self-learning function. Experiment results show that the method can achieve a better effect for phishing filter.