📰 NewsCheck AI

Streamlit app to analyze news articles for credibility, sentiment, clickbait, and more. 🔗 Live App: https://newscheck-ai.streamlit.app/

---

👥 Contributors

• Onyeka Nwokike — Team Lead / Topic Classification — https://github.com/Nwokike
• Stephen Ayankoso — Fake News Detection — https://github.com/Steve-ayan
• Cleiton Langa — Clickbait Detection — https://github.com/cleitonlanga
• Rivaldo — Sentiment Analysis — https://github.com/rivaldo56

---

📘 Project Overview

NewsCheck AI is a modular NLP project built with Streamlit. Each team member contributed a trained model that analyzes different aspects of news articles. Incomplete modules display as “Coming Soon.”

---

📁 Project Structure

newscheck-ai/
│
├── app.py
│
├── model_functions/
│   ├── topic.py
│   ├── fake.py
│   ├── sentiment.py
│   ├── clickbait.py
│   ├── bias.py         # Coming Soon
│   ├── summarizer.py   # Coming Soon
│   └── emotion.py      # Coming Soon
│
├── models/
│   ├── topic_model.pkl
│   ├── topic_vectorizer.pkl
│   ├── fake_news_model.pkl
│   ├── fake_news_vectorizer.pkl
│   ├── sentiment_model.pkl
│   ├── tfidf_vectorizer.pkl
│   ├── clickbait_model.pkl
│   ├── clickbait_vectorizer.pkl
│   └── ...
│
├── notebooks/
│   ├── topic_classification.ipynb
│   ├── fake_news_training.ipynb
│   ├── sentiment_training.ipynb
│   ├── clickbait_training.ipynb
│   └── ...
│
├── requirements.txt
├── .gitignore
└── README.md

---

🧠 Models

This table shows the high-level status of each module. See the "Data Sources" section below for the specific datasets and requirements for each model.

Feature	Function	Status
Topic Classification	predict_topic(text)	✅ Done
Fake News Detection	predict_fake(text)	✅ Done
Sentiment Analysis	predict_sentiment(text)	✅ Done
Clickbait Detection	is_clickbait(text)	✅ Done
Bias Detection	detect_bias(text)	🚧 Coming Soon
Extractive Summarizer	summarize(text)	🚧 Coming Soon
Emotion Detection	get_emotion(text)	🚧 Coming Soon

Data Sources & Contributor Guide

This is the official list of datasets for the project. Contributors working on a "Coming Soon" module should use the specified dataset and deliverable.

• Topic Classification (Done)

  • Dataset: News Category Dataset
  • Deliverable: A function predict_topic(text) in model_functions/topic.py that returns the predicted category as a string.

• Fake News Detection (Done)

  • Dataset: Fake and Real News Dataset
  • Deliverable: A function predict_fake(text) in model_functions/fake.py that returns a formatted string like "High Risk (Fake) - 92.51%".

• Sentiment Analysis (Done)

  • Dataset: Sentiment Analysis for Financial News
  • Deliverable: A function predict_sentiment(text) in model_functions/sentiment.py that returns "Positive," "Negative," or "Neutral."

• Clickbait Detection (Done)

  • Dataset: Clickbait Dataset
  • Deliverable: A function is_clickbait(headline_text) in model_functions/clickbait.py that returns "Yes" or "No".

• Bias Detection (Coming Soon)

  • Dataset: MBIC: A Media Bias Annotation Dataset
  • Deliverable: A function detect_bias(text) in model_functions/bias.py that returns "Biased" or "Neutral."

• Extractive Summarizer (Coming Soon)

  • Dataset: Newspaper Text Summarization CNN/DailyMail
  • Deliverable: A function summarize(text) in model_functions/summarizer.py that returns a single string containing the 3-sentence summary.

• Emotion Detection (Coming Soon)

  • Dataset: Emotions Dataset for NLP
  • Deliverable: A function get_emotion(text) in model_functions/emotion.py that returns the predicted emotion.

---

⚙️ Quick Start

1. Clone the repository:

   git clone https://github.com/Nwokike/newscheck-ai.git

2. Navigate to the directory:

   cd newscheck-ai

3. Create a virtual environment:

   python -m venv venv

4. Activate the environment:
   • macOS/Linux:

     source venv/bin/activate

   • Windows:

     venv\Scripts\activate

5. Install dependencies:

   pip install -r requirements.txt

6. Run the app:

   streamlit run app.py

---

🧩 Contributing

1. Fork the repo and create a new feature branch:

   git checkout -b feature/your-model

2. Add your files (see "Data Sources" section for model requirements):
   • Your training notebook $\rightarrow$ /notebooks
   • Your model + vectorizer $\rightarrow$ /models
   • Your inference script $\rightarrow$ /model_functions
3. Push your branch and open a Pull Request to main.