Fake News Analyzer

Final project for the Artificial Intelligence course [2025.1] at ICT - Unifesp

Special thanks to my partner, Luiza de Souza Ferreira for her contributions to this project

📚 About

Fake News Analyzer is a machine learning project designed to classify news articles as real or fake. It leverages traditional ML algorithms and modern NLP techniques, including SVM, Naive Bayes, KNN, and BERT-based transformers.

🚀 Features

• Preprocessing and cleaning of news datasets
• Training and evaluation of multiple ML models
• Support for both classical ML and transformer-based models

🛠️ Technologies

• Python 3.12+
• scikit-learn
• pandas, numpy
• TensorFlow & Hugging Face Transformers (for BERT)

📦 Project Structure

Fake-News-Analyzer/
│   .gitattributes
│   Apresentação Final - Fake News Analyzer.pdf
│   Fake News Analyzer.ipynb
│   pre-processed.csv
│   README.md
│   Relatório - Detecção de Fake News usando NLP.pdf
│
└───modelos/
    │   knn_model.pkl
    │   knn_vectorizer.pkl
    │   naive_bayes_model.pkl
    │   naive_bayes_vectorizer.pkl
    │   svm_model.pkl
    │   svm_vectorizer.pkl
    │
    └───bert/
        ├───bert_model/
        │       config.json
        │       tf_model.h5
        │
        └───bert_tokenizer/
                special_tokens_map.json
                tokenizer_config.json
                vocab.txt

⚡ Getting Started

1. Clone the repository:

   git clone https://github.com/JoaoPedroZampoli/Fake-News-Analyzer.git
   cd Fake-News-Analyzer

2. Install dependencies

   pip install numpy pandas matplotlib seaborn tensorflow transformers scikit-learn
   
   

3. Create the models

   • run Fake News Analyzer.ipynb cells with Python to create new models if wanted

📄 License

This project is for educational purposes.