DistilBERT Based Malicious URL Detection System
Aninda Kumar Sharma · 2024
The increasing threat of network information insecurity has led to a surge in malicious activities such as phishing, defacement, and malware distribution through deceptive URLs. This study proposes a novel approach to identify malicious URLs using the DistilBERT transformer model focusing on the improvement of detection accuracy. We built a complete data preprocessing pipeline which preprocessed the URL features utilizing big data technology to analyze URL patterns and anomalies. By doing so, the proposed system tends to categorize URLs into benign, defacement, phishing or malware groups thus enhancing detection accuracy significantly. The full dataset consisted of 651,191 URLs. Performance achieved by our model were 97.18% and 97.60% for accuracy and precision respectively. The experimental results show that the DistilBERT based model combined with our preprocessing techniques provides a robust and efficient solution for detecting malicious URLs. Any advancement in cybersecurity technology can substantially mitigate risks associated with malicious URLs, providing a reliable tool for protecting users and organizations. It is suggested that the proposed system may be considered as an optimized and friendly used solution for malicious URL detection.