# Malicious URL Detection using Machine Learning: A Survey

**Authors:** Doyen Sahoo, Chenghao Liu, Steven C. H. Hoi  
**arXiv:** [1701.07179](https://arxiv.org/abs/1701.07179)  
**Date:** 2017-01-25  
**Summary page:** https://www.summarizepaper.com/en/paper/1701.07179v1/

## Key points

- Malicious URLs pose a significant threat to cybersecurity
- They host unsolicited content such as spam, phishing attempts, and drive-by exploits
- Financial impact of these threats amounts to billions of dollars annually
- Traditional detection methods rely on blacklists, which have limitations
- Machine learning techniques are increasingly being explored for detecting malicious URLs
- The article presents the formal formulation of Malicious URL Detection as a machine learning task
- Various contributions from literature studies addressing different dimensions of this problem are categorized and reviewed
- Intended for machine learning researchers, engineers, professionals, and practitioners in the cybersecurity industry
- Discusses practical issues in system design related to malicious URL detection
- Highlights open research challenges and important directions for future research.

## Summary

The article titled "Malicious URL Detection using Machine Learning: A Survey" provides a comprehensive survey and structural understanding of the techniques used for detecting malicious URLs through machine learning. The authors emphasize that malicious URLs pose a significant threat to cybersecurity, as they host unsolicited content such as spam, phishing attempts, and drive-by exploits. These malicious websites lure unsuspecting users into becoming victims of scams, resulting in monetary loss, theft of private information, and malware installation. The financial impact of these threats amounts to billions of dollars annually. Traditionally, the detection of malicious URLs has relied on the usage of blacklists. However, blacklists have limitations as they cannot be exhaustive and fail to detect newly generated malicious URLs. To overcome these challenges and improve the generality of malicious URL detectors, researchers have increasingly explored machine learning techniques. The article presents the formal formulation of Malicious URL Detection as a machine learning task and categorizes and reviews various contributions from literature studies addressing different dimensions of this problem. These dimensions include feature representation and algorithm design among others. This survey is not only intended for machine learning researchers and engineers in academia but also for professionals and practitioners in the cybersecurity industry. It aims to help them understand the state-of-the-art techniques in detecting malicious URLs using machine learning and facilitate their own research and practical applications. Additionally, the article discusses practical issues in system design related to malicious URL detection, open research challenges that need further exploration, and important directions for future research. Overall, this comprehensive survey provides valuable insights into the field of Malicious URL Detection using machine learning techniques. It highlights the importance of timely threat detection to mitigate potential losses caused by these cybersecurity threats while also providing guidance for future advancements in this area.