QuickCharNet

QuickCharNet: Enhancing Search Engine Optimization Using Efficient Character-Level Convolutional Network for non-optimal URL Detection

Introduction

This project is part of Quick Char Net. This project explores the use of deep learning techniques, specifically character-level Convolutional Neural Networks (CNNs), for URL classification. The goal is to classify URLs based on their appearance, which can be used for SEO optimization.

Project Structure

• datasets/: Contains datasets used for training and evaluation.
• FindBestModel/: Contains Jupyter notebooks for testing and identifying the best model.
• TestsOnGrambedding/: Contains Jupyter notebooks for testing and comparing the model against state-of-the-art models using the Grambedding dataset.
• TestsOnMaliciousURLs/: Contains Jupyter notebooks for testing and comparing the model against state-of-the-art models using the Malicious URLs dataset.
• TestsOnPhishStorm/: Contains Jupyter notebooks for testing and comparing the model against state-of-the-art models using the PhishStorm dataset.
• TestsOnSpamURLs/: Contains Jupyter notebooks for testing and comparing the model against state-of-the-art models using the Spam dataset.
• Visualizations/: Python scripts to show how models work include evaluations and how they create their outputs.

Methodology

QuickCharNet Architecture

<img alt="The model architecture" src="Model1Architecture2.jpg">

A brief explanation of the seed (here, 911):

This model, like many others, works best with a specific initial state. Experimentally, it performed best using the fixed initial state with seed 911, although you might, theoretically or experimentally, find a better fixed or random initial state. The initial state 911 worked well with various datasets and learning conditions—not just under a single human-engineered condition.

Dependencies

Note: This project is no longer maintained, as it has not received any citations. It was developed as part of a research paper. To get the most out of it, please do not rely solely on the requirements.txt file—update the dependencies to their latest compatible versions.

requirements.txt: Use this file to install required packages.

Citation

@ARTICLE{rastakhiz2024quick,
  author={Rastakhiz, Fardin and Eftekhari, Mahdi and Vahdati, Sahar},
  journal={IEEE Access}, 
  title={QuickCharNet: An Efficient URL Classification Framework for Enhanced Search Engine Optimization}, 
  year={2024},
  volume={12},
  number={},
  pages={156965-156979},
  keywords={Search engines;Uniform resource locators;Optimization;Machine learning;Web pages;Text categorization;Accuracy;Semantics;Analytical models;Phishing;Convolutional neural networks;Deep learning;Ranking (statistics);Convolutional neural networks;deep learning;malicious URL detection;text classification;web page ranking},
  doi={10.1109/ACCESS.2024.3484578}}

License

For detailed licensing information, please see the LICENSE file.

QuickCharNet, Quick Char Net, Convolutional Neural Network, URL classification, Spam Detection, Search Engine Optimization

Support & Contributions

Have questions, feedback, or ideas to improve this project? Reach out via:

• Telegram: @FardinRastakhiz