# Amazon Product Hierarchy Dataset

This dataset contains Amazon product hierarchical relationships for training hierarchy-aware transformers.

## Dataset Structure

The dataset contains three splits:
- Train: 3 examples
- Validation: 3 examples
- Test: 3 examples

Each example contains:
- `child`: Text description of a product
- `parent`: Text description of the product's category
- `negative`: Text description of a negative (non-parent) category

## Usage

```python
from datasets import load_dataset

# Load the dataset
dataset = load_dataset("Jinrui/amazon-product-hierarchy")

# Example usage
example = dataset['train'][0]
print(f"Child: {example['child']}")
print(f"Parent: {example['parent']}")
print(f"Negative: {example['negative']}")
```

## Dataset Creation

This dataset was created by processing Amazon Beauty product metadata and reviews, extracting product categories and their hierarchical relationships.