Detection and Analysis of Malicious URLs

Abstract

Social Network Sites(SNS) is the soul of the Internet. It has become a global phenomenon with enormous social as well as economic importance within a few years of their launch. Because of larger user space SNS has become popular day by day. Information exploitation popularity in SNS has attracted not only novice users but also spammers. In SNS spammers are using evolving technology and they safely trading their illegal activities by phishing through e-mails, Social Reverse Engineering(SRE), by posting some incite messages. The novice users often becomes victim to these malicious activity which impacts them both socially and economically. The study show that because of this illegal activity the SNS organisers and users are loosing $2 million for three months. In this thesis we exploited the security gap that many popular SNS services like Twitter, Facebook do not provide to its users. We have collected a large scale of long URLs and short URLs from multiple sources of SNS which are checked against malicious and non-malicious detectors and we analyse their features to classify the URLs. Our result shows that Naïve Bayes classifier performs better then other classifier algorithm with accuracy 95.4% [math mode missing closing $]