PhishBuster: An Intelligent Web-Based Tool for Real-Time Malicious URL Detection in Small Businesses

Abstract

In light of the ongoing digital transformation, small and medium-sized enterprises (SMEs) in Peru are becoming increasingly susceptible to phishing attacks, which threaten both operational continuity and the protection of sensitive data. To tackle this issue, this study introduces a smart web-based solution designed to detect malicious URLs by leveraging machine learning (ML) techniques. The main objective of this study is to develop and evaluate a machine learning-based browser extension capable of accurately identifying phishing URLs in realtime scenarios. The system was assessed using three classification algorithms—XGBoost, LightGBM, and Random Forest—trained on publicly available datasets from PhishTank and PhishStorm. The performance of each model was evaluated using key metrics, including accuracy, precision, recall, specificity, F1-score, receiver operating characteristic curve (ROC), and the area under the ROC curve (AUC). Among the tested models, XGBoost achieved the highest performance, recording an AUC of 0.99 and an accuracy of 94.6%. The tool proved effective in identifying phishing links, especially by reducing the rate of false negatives, which is crucial for real-time threat prevention. In addition, a continuity strategy was developed to ensure smooth integration into the digital environments of SMEs. This proposed solution stands out for its ease of deployment, scalability, and efficiency, offering a meaningful contribution to improving cybersecurity and strengthening the digital resilience of Peru’s SME sector.