Enhanced Framework for Detecting Malicious HTTP Redirections with Supporting Classifier

Abstract

Malicious URL detection has become increasingly difficult due to the evolution of phishing campaigns and efforts to avoid weakening blacklists. The existing state of cybercrime has allowed pirates to host campaigns with smaller lifespan, which reduces the efficacy of the backlist. At the same time, standard supervised learning algorithms are known to generalize in specific patterns observed in the training data, which makes them a better alternative against piracy campaigns. However the highly dynamic environment of these campaigns requires models updated frequently, which poses new challenge as most learning algorithms are too computationally require exclusive retraining. This paper surveys two contributions. Firstly it discusses the problems associated with Malicious URL and there propagation mechanism. Secondly, it provides method to detect and distinguish Malicious URL by analyzing them. For analysis Recall, Precision and F-measures matrices are used.