Sentiment Pipeline with OpenAI

This project provides a modular pipeline for classifying text data into sentiments (positive, negative, neutral) using the OpenAI API.
It is adaptable to multiple domains such as product reviews, news headlines, or any short text format.

Built with Python, the system supports both generative AI-based classification and traditional evaluation via scikit-learn.

---

💡 Project Objectives

• Apply Generative AI to classify text data sentiment.
• Support multiple domains via reusable configurations (nyt, reviews, etc.).
• Enable clean modularization: preprocessing → classification → evaluation.
• Provide a CLI interface for production-ready usage.

---

🚀 Technologies Used

• Python 3.8+
• OpenAI API (gpt-3.5-turbo)
• Pandas – data manipulation
• scikit-learn – evaluation metrics
• dotenv – API key management
• argparse – CLI interface
• Jupyter Notebook – EDA (optional)

---

📁 Project Structure

project-root/
├── data/
│   ├── raw/               # Unprocessed input data
│   └── processed/         # Cleaned and classified output
│
├── notebooks/             # EDA and insights (optional)
│
├── src/
│   ├── config/            # Modular configuration per domain
│   │   ├── base_config.py
│   │   ├── reviews_config.py
│   │   └── nyt_config.py
│   ├── data_preprocessing.py
│   ├── classifier.py
│   ├── evaluation.py
│   └── cli_runner.py      # Main CLI script
│
├── .env                   # API key (not committed)
├── requirements.txt
└── README.md

---

⚙️ How to Run the Project

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/Sentiment-Pipeline-With-OpenAI.git
cd Sentiment-Pipeline-With-OpenAI

2. Set up environment

python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Add your OpenAI API Key

Create a file called .env in the root directory:

OPENAI_API_KEY=your-key-here

---

🧐 Running the Pipeline

python src/cli_runner.py --domain reviews

Available domains:

• reviews – product review sentiment analysis
• nyt – sentiment classification of NYT headlines

---

📦 Input Requirements

For reviews:

• File: data/raw/reviews.csv
• Column required: cleaned_text

For nyt:

• File: data/raw/nyt_articles.json
• Field required: headline

---

📊 Output

• Classified data in: data/processed/<domain>_classified.csv
• If has_labels=True in config, an evaluation report is printed.

---

✨ Coming Soon

• Support for local models (e.g., BERT)
• Streamlit frontend
• Logging and metrics dashboard
• Custom prompt tuning via config

---

📄 License

MIT License © 2025 William Yohei