Replication Data & Source Code: A Lightweight Browser-Based Phishing Detection System using Optimized Homoglyph Mapping

This repository contains the replication data, source code, and synthetic datasets associated with the research paper: "A Lightweight Browser-Based Phishing Detection System using Optimized Homoglyph Mapping for Real-Time Protection." The artifacts provided here allow researchers to reproduce the experiments described in the paper, specifically evaluating the detection accuracy against sophisticated Whole-Script Homograph attacks that often bypass standard browser protections. Repository Contents: Source Code (Backend & Extension): The implementation of the "Thin-Client, Heavy-Backend" architecture using Python Flask and a lightweight browser extension. Optimized Homoglyph Map (tdv_homoglyph_map.txt): A custom, resilience-optimized map developed to bridge the gap between standard Unicode tables and browser rendering behaviors. Synthetic Phishing Dataset: A collection of 7,000 generated malicious domains covering five attack vectors: Single Mixed-Script Homographs Diacritic Substitutions SLD/TLD Script Mismatches Whole-Script Homographs (The "Zero-Day" Scenario) Typosquatting Benign Dataset: A pre-configured subset of the Tranco Top 1 Million list used for false-positive testing. Testing Scripts: Python automation scripts used to benchmark latency, memory usage, and detection accuracy (accuracy_testing.py, real_domain_test.py). Note: For the latest version of the software and installation guides, please visit the active GitHub repository linked in the paper.