Phroura

Phroura is a master's-level computer science project developed in fulfillment of the requirements for DLMCSPCSP01 - Project: Computer Science. The project focuses on phishing URL detection using lexical feature engineering, classical machine learning, and a lightweight full-stack interface.

Project Goal

The goal of this project is to design, implement, and evaluate a phishing detection platform that can classify suspicious URLs in real time using lightweight, explainable methods.

Planned Repository Structure

• app/ - Streamlit interface and user-facing application code
• data/ - dataset files and dataset notes
• models/ - trained model artifacts and related metadata
• notebooks/ - exploratory analysis and experiments
• src/ - core source code for features, training, evaluation, and inference
• tests/ - tests for feature extraction, inference, and other core logic
• docs/ - private planning and learning documents

Planned Core Workflows

1. Select and document phishing and benign URL datasets.
2. Extract lexical and structural features from URLs.
3. Compare heuristic and machine learning phishing detectors.
4. Integrate the selected model into a usable application.
5. Evaluate the system and document the results in the final report.

Status

Sprint 2 now has a working lexical feature extraction module, representative tests, and sample feature output generated from raw URLs.