A Framework to Detect Compromised Websites Using Link Structure Anomalies

Abstract

Compromised or malicious websites are a serious threat to cyber security. Malicious users prefer to do malicious activities like phishing, spamming, etc., using compromised websites because they can mask their original identities behind these compromised sites. Compromised websites are more difficult to detect than malicious websites because compromised websites work in masquerade mode. This is one of the main reasons for us to take this topic as our research. This paper introduces the related work first and then introduces a framework to detect compromised websites using link structure analysis. One of the most popular link structures based ranking algorithm used by the Google search engine algorithm called PageRank, is implemented in our experiment using the Java language, computation is done before a website is compromised and after a website is compromised and the results are compared. The results show that when a website is compromised, its PageRank can go do down to indicate that this website is compromised.