🚀 Emotion Analysis for Vietnamese Text

This project implements a deep learning model for emotion analysis in Vietnamese text, with both training pipeline and API endpoints for real-world applications.

📸 Screenshots

📌 Result analyze

📌 Comment Relevance

🔌 Pipeline Diagrams

📈 Training Pipeline

📈 API Pipeline

📈 Model Evaluation Results

• Accuracy: 62,2%
• F1-Score: 0.62
• Confusion Matrix Results are available in reports/model_evaluation.ipynb

Detailed evaluation metrics for each emotion:

Emotion	Precision	Recall	F1-Score	Support
Anger	0.41	0.35	0.38	40
Disgust	0.59	0.61	0.60	132
Enjoyment	0.70	0.67	0.68	193
Fear	0.63	0.67	0.65	46
Sadness	0.72	0.61	0.66	116
Surprise	0.65	0.59	0.62	37
Other	0.54	0.64	0.58	119

📦 Installation

Option 1: Use venv

1. Clone the repository:

git clone https://github.com/PhucHau0310/Emotion-Analysis.git
cd EmotionAnalyze

2. Create and activate virtual environment:

python -m venv .venv
.\.venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Install Graphviz (for pipeline visualization):
   • Download from: https://graphviz.org/download/
   • Add to system PATH

Option 2: Use docker-compose

cd root project
docker-compose up --build

🧪 Usage

Training

cd be/src
python scripts/train.py

Running API Server

cd be
uvicorn src.main:app --reload

API Endpoints

1. YouTube Comments Analysis:

POST /analyze
{
    "link": "https://www.youtube.com/watch?v=VIDEO_ID"
}

2. Text Analysis:

POST /predict
{
    "text": "Tôi rất vui khi gặp lại bạn!"
}

📁 Project Structure

EmotionAnalyze/
├── be/
│   ├── src/
│   │   ├── models/
│   │   ├── utils/
│   │   ├── scripts/
│   │   └── config/
│   ├── reports/
│   ├── notebooks/
│   └── tests/
├── requirements.txt
├── fe/ (next.js)
├── docker-compose.yml
└── README.md

🛠️ Environment Variables

Create a .env file in the root directory be:

YOUTUBE_API_KEY=your_api_key_here

🙋 Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

👤 Authors

• Nguyễn Phúc Hậu - Initial work
• Email: haunhpr024@gmail.com
• Github: PhucHau0310

❓ Acknowledgments

• PhoBERT for providing pre-trained Vietnamese language model
• FastAPI for the web framework
• YouTube API for comment extraction capabilities