Build an intelligent support ticket classification system using Large Language Models (LLMs) with prompt engineering, comparing zero-shot vs few-shot learning approaches.
Synthetic Support Tickets (20 samples for demonstration)
Categories:
- Billing (payment issues, refunds, charges)
- Technical (bugs, crashes, errors)
- Account (login, password, settings)
- Shipping (delivery, tracking, damaged items)
- Feature Request (new features, integrations)
- Inquiry (general questions, information)
- Security (suspicious activity, 2FA)
- Urgent (immediate attention needed)
| Technology | Purpose | Free Tier |
|---|---|---|
| Python 3.8+ | Programming | ✅ |
| pandas/numpy | Data processing | ✅ |
| Google Gemini | LLM API (recommended) | ✅ 60 req/min |
# Base packages
pip install pandas numpy scikit-learn
pip install google-generativeai # For Gemini
results.csv- All predictionsmetrics.json- Performance metrics