A website credibility assessment scheme based on page association

The credibility of websites is an important factor to prevent malicious attacks such as phishing. These attacks cause huge economic losses, for example attacks to online transaction systems. Most of the existing page-rating solutions, such as PageRank and Alexa Rank, are not designed for detecting malicious websites. The main goal of these solutions is to reflect the popularity and relevance of the websites, which might be manipulated by attackers. Other security-oriented rating schemes, e.g., black/white listed based, voting-based and pagesimilarity- based mechanisms, are limited in the accuracy for new pages, bias in recommendation and low efficiency. To balance the user experience and detection accuracy, inspired by the basic idea of PageRank, we developed a website credibility assessment algorithm based on page association. We prototyped our algorithm and developed a website assessment extension for the Safari browser. The experiment results showed that our method is accurate and effective in assessing websites for threats from phishing with a low performance overhead.