Detecting malicious short URLs on Twitter

Short URLs (Uniform Resource Locators) have gained immense popularity especially in Online Social Networks (OSNs), blogs, and messages. Short URLs are used to avoid sharing overly long URLs and save limited text space in messages or tweets. Significant numbers of URLs shared in the Online Social Networks are shortened URLs. Despite of its potential benefits from genuine usage, attackers use shortened URLs to hide the malicious URLs, which direct users to malicious pages. Although, OSN service providers and URL shortening services utilize certain detection mechanisms to prevent malicious URLs from being shortened, research has found that they fail to do so effectively. These malicious URLs are found to propagate through OSNs. In this paper, we propose a mechanism to develop a machine learning classifier to detect malicious short URLs with visible content features, tweet context, and social features from one popular Online Social Network Twitter.