FAKE WEBSITE DETECTION USING MACHINE LEARNING ALGORITHMS

Computer Engineering

The rise of deceptive online platforms such as phishing and fake websites poses a critical threat to digital security and personal privacy. This paper presents a comprehensive study on the implementation of machine learning algorithms for detecting fake websites based on URL and metadata features. Using a dataset derived from real-world phishing cases, we evaluate the performance of multiple supervised learning models—specifically Random Forest, LightGBM, and XGBoost—within a Java-based framework powered by the Weka library. These algorithms analyze structural and lexical patterns in website URLs to distinguish between legitimate and malicious domains. The system is designed for integration into enterprise-grade cybersecurity tools, offering rapid classification, scalability, and adaptability to new threats. Evaluation results demonstrate that ensemble models, particularly Random Forest, provide superior accuracy and reliability, making them suitable for real-time deployment in browsers, email gateways, and network firewalls. While the study confirms the feasibility and efficiency of Java-based ML solutions for web threat detection, it also addresses challenges such as evolving phishing techniques, feature selection, and data quality. Through empirical analysis and performance benchmarking, this research offers practical insights and a robust foundation for future advancements in automated web threat detection and secure web browsing technologies.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.