Login Search
FMDB Transactions on Sustainable Computing Systems Submit Paper
    Crossref logo

    Archiving

    Portico

    Fernando Martins De Bulhão is a member of

    Crossref logo

Intelligent Hybrid Machine Learning System for Phishing Website Detection Using URL and Content Based Features

Authors:
Miracle Kelechi Chibueze

Addresses:
Department of Computer and Information Sciences, Northumbria University, Newcastle Upon Tyne, England, United Kingdom.

Abstract:

Financial fraud and password theft from phishing attempts cost people and organizations billions of dollars annually. Blocklists and basic heuristics struggle to identify advanced phishing sites with real URLs and high-quality content. This study developed a hybrid machine learning system for real-time phishing detection that leverages URL- and content-based features. The overall technique used PhiUSIIL, 235,795 URLs, and 56 features for data pretreatment, feature engineering, and model tuning. Stratified cross-validation trained and compared Random Forest and XGBoost ensemble learning algorithms to evaluate their performance. The hybrid system examined 25 URLs (lexical patterns, domain properties, and structural properties) and 29 content (HTML structure, JavaScript behaviour, form elements, and metadata). XGBoost beat Random Forest with 99.98 test accuracy and false-positive and false-negative error rates of 0.0049% and 0.0462%, respectively. Analysis of significance. HasSocialNet alone has 39.54% discriminative power, whereas content-based qualities have 65.1%. Comparing the hybrid model to URL-only baselines improved error by 60%. The computational performance analysis demonstrated real-time performance with an inference delay of less than 20 milliseconds and efficient batch processing. The six-dimensional usability test averaged 9.2/10. This study suggests that hybrid machine learning algorithms can detect phishing, explain the discriminative feature hierarchy, and solve cybersecurity problems.

Keywords: Phishing Detection; Financial Fraud; Password Theft; Hybrid Machine Learning; Feature Engineering; Model Tuning; Stratified Cross-Validation; Random Forest; Ensemble Learning; Lexical Patterns.

Received on: 20/04/2025, Revised on: 17/06/2025, Accepted on: 22/09/2025, Published on: 12/06/2026

DOI: 10.69888/FTSCS.2026.000686

FMDB Transactions on Sustainable Computing Systems, 2026 Vol. 4 No. 2, Pages: 106-134

  • Views : 39
  • Downloads : 8
Download PDF