Machine Learning Approaches for Malicious URL Detection: A Literature Survey

Machine Learning Approaches for Malicious URL Detection: A Literature Survey

Meenu S Nair
Department of Computer Science (Cyber Security)

Vimal Jyothi Engineering College Chemperi, Kannur Email:meenusnair692@gmail.com

Sanjay S Kumar
Department of Computer Science (Cyber Security)

Vimal Jyothi Engineering College Chemperi, Kannur Email:sanjaysk.7704@gmail.com

Muhammed Hawaz PK

Department of Computer Science (Cyber Security)

Vimal Jyothi Engineering College Chemperi, Kannur Email:muhammedhawaz1222@gmail.com

Akshay
Department of Computer Science (Cyber Security)

Vimal Jyothi Engineering College

Chemperi, Kannur Email:akshaysajeevan557@gmail.com

Mr.Rasheed Ahamed Azad V
Assistant Professor

Department of Computer Science (Cyber Security)

Vimal Jyothi Engineering College Chemperi, Kannur Email:rasheedklpm@vjec.ac.in

Abstract—As cyber threats keep getting more advanced, mali- cious URLs have become a huge problem, causing data breaches, malware infections and financial losses. Hackers use tricky meth- ods, like sneaking harmful links into emails, social media and even legit-looking websites. Traditional blacklist-based detection isn’t very reliable since it needs constant updates and often misses new threats. To tackle this, machine learning provides a smarter approach by analyzing patterns and behaviors of malicious URLs. Algorithms like SVM, Random Forest, Logistic Regression and deep learning models like XGBoost and MLPs have shown really good accuracy in detecting harmful links. But there are still challenges, like dealing with imbalanced datasets, needing high-quality data and handling the high computational costs. This survey looks into different machine learning-based detection techniques, their strengths and weaknesses and why ongoing improvements are necessary to stay ahead of evolving cyber threats.

Index Terms—Machine Learning , Malicious URL, Cyberse- curity , Malware Detection , LSTM , Kaggle , Safe Browsing.