An content-analysis based large scale Anti-Phishing Gateway

In this paper, we introduce the design, implementation and performance characteristics of BUPT-APG, our novel, content-analysis based phishing detection system. Unlike previous works in this field, APG focuses on the path between the user´s browser and the interacted web server. We, meanwhile, introduce a novel algorithm which based on an adjusted cosine similarity to calculate the similarity between the template repository and target pages. Leveraging this appropriate algorithm of our phishing detection techniques, the system has achieved more then 98% accuracies and 0.053% recall with a satisfied processing speed.