This project is fully reproducible and designed to run directly in the Kaggle Notebook environment.

Access the public Kaggle notebook here:

👉 Kaggle Notebook
https://www.kaggle.com/code/thanhtamdo91/openstax-2026-rice-datathon-mapping-standards-i

1. Click “Copy & Edit” (top right)
2. This will create your own editable copy in your Kaggle workspace

1. In the notebook, click Settings
2. Set:
   • Accelerator: GPU
   • GPU Type: T4 ×2 (recommended)
3. Save settings

This project uses the Gemini API for data augmentation.

1. Go to Kaggle → Settings → Secrets
2. Add a new secret:
   • Name: GEMINI_API_KEY
   • Value: your Gemini API key
3. The notebook automatically loads it using kaggle_secrets

If the API key is not provided, the notebook will still run using the base dataset.

1. Click Run All
2. The notebook will:
   • Parse and flatten the OpenStax JSON data
   • Perform EDA
   • Generate embeddings and FAISS index
   • Run semantic retrieval + re-ranking
   • Output evaluation metrics and visualizations

• GPU Enabled: ~10–15 minutes
• CPU Only: Not recommended (embedding + FAISS steps are slow)

• Evaluation metrics (Hit@K, F1, Hamming Loss)
• Retrieval performance analysis
• Error distribution across standard codes
• Augmented training artifacts (optional)

• No local setup is required
• No additional dependencies need to be installed
• All datasets are loaded directly from Kaggle inputs