Logo image
Back
Phishing Defense: An ML-Based URL Detection System with Real-World Deployment
Conference proceeding   Peer reviewed

Phishing Defense: An ML-Based URL Detection System with Real-World Deployment

ABM Kamrul Islam Riad, Md Reazul Hassan Rizvi, Abdullah Al Mamun, Shakil Miah, Md Abdul Barek, Yasmeen Rawajfih, Muhammad Umair Khan, Hossain Shahriar and Alfredo Cuzzocrea
IEEE International Conference on Big Data, (2025), pp.4353-4359
IEEE International Conference on Big Data (BigData) (Macau, China , 12/08/2025–12/11/2025)
03/2026
DOI: https://doi.org/10.1109/BigData66926.2025.11402049

Metrics

1 Record Views

Abstract

Phishing detection Random Forest UN SDG 9 URL analysis Cybersecurity Machine Learning
Phishing attacks are a growing concern in cybersecurity, as they exploit human vulnerabilities through deceptive techniques to gain unauthorized access to sensitive information. Traditional signature-based detection tools often fail to identify newly created phishing websites. To address this, machine learning (ML) methods offer more accurate and adaptive detection. The use of many features in a model makes computations heavier, which extends training time while possibly reducing performance in real-time systems. Feature selection techniques help reduce computational overhead by identifying the most relevant attributes without compromising accuracy. The study assesses the performance of four classification algorithms, including Logistic Regression (LR), Decision Tree (DT), Naïve Bayes (NB), and Random Forest (RF), on a dataset that contains an equal number of phishing and benign URLs using 25 attributes to describe each instance. The study demonstrates that the Random Forest classifier delivers better performance than its competitors having 93.83% accuracy that remains stable both before and after feature selection with faster training times. We constructed an interactive web platform that integrates the optimized Random Forest model for instant phishing URL detection. Additionally, we demonstrate that efficient feature selection enables the development of precise, lightweight detection systems that both strengthen cybersecurity defense and support digital innovation.

Details

Logo image