Detection of Potentially Hazardous Websites Utilizing Hidden Markov Models

Malicious URLs pose a significant cybersecurity threat due to spread malicious software and obtain personal data, and initiate phishing attempts.Traditional techniques for identifying fraudulent URLs include blacklists and heuristics, are losing their effectiveness as new evasion strategies are created by attackers.A study titled "Malicious URL Detection based on Machine Learning" introduces a novel approach using Hidden Markov Models (HMM) to quantify and predict the behavior of Malicious Web Services.This approach not only measures but also predicts the response time of these services, allowing for a quantitative ranking rather than a qualitative assessment.The proposed methodology aims to automatically select the most reliable Malicious Web Service by considering metrics like system predictability and response time variability.The experimental data shows that the new model performs much better than the baseline.Comparing the recommended model to the state-of-the-art, the results show that it is more stable and has accuracy and recall rates of up to 99.2% and 99.16%, respectively.