gatech-scheller-ai-in-business/amazon-products

Amazon Products Sample Dataset

A curated sample of 2,000 popular products from the Amazon Reviews 2023 dataset, designed for educational use in building RAG (Retrieval-Augmented Generation) systems and shopping agents.

Dataset Description

This dataset contains product metadata across 4 categories:

• —Electronics (500 products)
• —Video Games (500 products)
• —Books (500 products)
• —Home & Kitchen (500 products)

Products were filtered to include only those with 500+ reviews, ensuring recognizable, well-documented items.

Dataset Structure

Each product includes:

• —parent_asin: Unique product identifier
• —title: Product title
• —description: Product description (list of strings)
• —price: Listed price (string)
• —price_numeric: Price as float
• —average_rating: Average star rating (1-5)
• —rating_number: Number of reviews
• —features: Product features (list)
• —categories: Category hierarchy (list)
• —category: Simplified category label
• —store: Store/brand name
• —images: Product images (list of dicts)
• —main_category: Original main category

Usage

from datasets import load_dataset

dataset = load_dataset("gatech-scheller-ai-in-business/amazon-products")
df = dataset['train'].to_pandas()

# Filter by category
electronics = df[df['category'] == 'Electronics']

Source & Citation

This dataset is derived from the Amazon Reviews 2023 dataset compiled by the McAuley Lab at UC San Diego.

Original Dataset: McAuley-Lab/Amazon-Reviews-2023

Project Page: https://amazon-reviews-2023.github.io/

If you use this data, please cite the original work:

@article{hou2024bridging,
  title={Bridging Language and Items for Retrieval and Recommendation},
  author={Hou, Yupeng and Li, Jiacheng and He, Zhankui and Yan, An and Chen, Xiusi and McAuley, Julian},
  journal={arXiv preprint arXiv:2403.03952},
  year={2024}
}

License

This dataset follows the licensing terms of the original Amazon Reviews 2023 dataset. It is intended for research and educational purposes only.

Intended Use

This dataset was created for the Georgia Tech Scheller College of Business course on AI in Business, specifically for assignments involving:

• —Building RAG systems for product search
• —Creating tool-using LLM agents
• —E-commerce recommendation systems

Limitations

• —Products are filtered by review count (500+ reviews), which may introduce popularity bias
• —Prices and availability may be outdated
• —Not intended for production e-commerce applications