Towards Lightweight URL-Based Phishing Detection

Nowadays, the majority of everyday computing devices, irrespective their size and operating system, allow access to information online services through web browsers. However, pervasiveness browsing in our daily life does not come without security risks. This widespread practice combination with users’ low situational awareness against cyber attacks, exposes them a variety threats, such as phishing, malware profiling. Phishing attacks can compromise target, individual or enterprise, social interaction alone. Moreover, current threat landscape phishing typically serve an attack vector initial step more complex campaign. To make matters worse, past work has demonstrated inability denylists, which are default countermeasure, protect users from dynamic nature URLs. In this context, uses supervised machine learning block based on novel features that extracted solely URL. We evaluate performance over time dataset consists active compare it Google Safe Browsing (GSB), i.e., control most popular find outperforms GSB all experiments, well performs even URLs one year after model’s training.