---
task_categories:
- text-classification
- feature-extraction
- sentence-similarity
language:
- en
tags:
- e-commerce
- products
- amazon
size_categories:
- 100K<n<1M
license: mit
---

# Dataset Card for Amazon Products 2023

## Dataset Summary

This dataset contains product metadata from Amazon, filtered to include only products that became available in 2023. The dataset is intended for use in semantic search applications and includes a variety of product categories.

- **Number of Rows:** 117,243
- **Number of Columns:** 15

## Data Source

The data is sourced from [Amazon Reviews 2023](https://amazon-reviews-2023.github.io/). 
It includes product information across multiple categories, with additional embeddings. 

Embeddings were created from the **title** + **description** using the **text-embedding-3-small** model.

## Dataset Structure

### Number of Products by Filename

|    | filename                                |   product_count |
|---:|:----------------------------------------|----------------:|
|  0 | meta_Amazon_Fashion                     |             470 |
|  1 | meta_Appliances                         |             573 |
|  2 | meta_Arts_Crafts_and_Sewing             |            2948 |
|  3 | meta_Automotive                         |            7161 |
|  4 | meta_Baby_Products                      |             526 |
|  5 | meta_Beauty_and_Personal_Care           |            1402 |
|  6 | meta_Books                              |               2 |
|  7 | meta_CDs_and_Vinyl                      |            1319 |
|  8 | meta_Cell_Phones_and_Accessories        |            5062 |
|  9 | meta_Clothing_Shoes_and_Jewelry         |           41777 |
| 10 | meta_Digital_Music                      |              56 |
| 11 | meta_Electronics                        |            7681 |
| 12 | meta_Gift_Cards                         |               8 |
| 13 | meta_Grocery_and_Gourmet_Food           |              96 |
| 14 | meta_Handmade_Products                  |            1018 |
| 15 | meta_Health_and_Household               |            4760 |
| 16 | meta_Health_and_Personal_Care           |              93 |
| 17 | meta_Home_and_Kitchen                   |           17326 |
| 18 | meta_Industrial_and_Scientific          |            1216 |
| 19 | meta_Magazine_Subscriptions             |               3 |
| 20 | meta_Musical_Instruments                |             639 |
| 21 | meta_Office_Products                    |            3545 |
| 22 | meta_Patio_Lawn_and_Garden              |            3075 |
| 23 | meta_Pet_Supplies                       |            2742 |
| 24 | meta_Software                           |             157 |
| 25 | meta_Sports_and_Outdoors                |            6343 |
| 26 | meta_Tools_and_Home_Improvement         |            4776 |
| 27 | meta_Toys_and_Games                     |            1367 |
| 28 | meta_Unknown                            |             541 |
| 29 | meta_Video_Games                        |             561 |


### Columns

- **parent_asin (str):** Unique identifier for the product.
- **date_first_available (datetime64[ns]):** The date when the product first became available.
- **title (str):** Title of the product.
- **description (str):** Description of the product.
- **filename (str):** Filename associated with the product metadata.
- **main_category (str):** Main category of the product.
- **categories (List[str]):** Subcategories of the product.
- **store (str):** Store information for the product.
- **average_rating (float64):** Average rating of the product.
- **rating_number (float64):** Number of ratings for the product.
- **price (float64):** Price of the product.
- **features (List[str]):** Features of the product.
- **details (str):** Additional details of the product. The string is JSON serializable.
- **embeddings (List[float64]):** Embeddings generated for the product using **text-embedding-3-small** model.
- **image (str):** URL of the product image.

### Missing Values

- **main_category:** 24,805 missing values
- **store:** 253 missing values
- **rating_number:** 6 missing values
- **price:** 35,869 missing values

### Sample Data

```json
[
  {
    "parent_asin": "B000044U2O",
    "date_first_available": "2023-04-29T00:00:00",
    "title": "Anomie & Bonhomie",
    "description": "Amazon.com Fans of Scritti Politti's synth-pop-funk masterpiece Cupid & Psyche 85 may be shocked by how far afield Scritti mastermind Green Gartside has gone since then. Anomie & Bonhomie, his return to recording after a decadelong absence, ranges from guest shots by rappers and funksters such as Mos Def and Me'Shell Ndegeocello to Foo Fighters tributes. Gartside's trademark breathy vocals and spot-on melodicism do find their places here, but are often forced to make way for other influences. Neither a total success nor a total failure, Anomie does display a spark that makes one hope that Gartside doesn't wait so long to record again. --Rickey Wright",
    "filename": "meta_Digital_Music",
    "main_category": "Digital Music",
    "categories": [],
    "store": "Scritti Politti Format: Audio CD",
    "average_rating": 4.2,
    "rating_number": 56.0,
    "price": null,
    "features": [],
    "details": "{'Date First Available': 'April 29, 2023'}",
    "embeddings": [],
    "image": "https://m.media-amazon.com/images/I/41T618NE88L.jpg"
  },
  ...
]
```

### Usage
This dataset can be used for various applications, including:

- Semantic Search: Utilizing the embeddings to find similar products based on textual descriptions.
- Product Recommendation: Enhancing recommendation systems with detailed product metadata.

### Citation

```bibtex
@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}
}
```

### Contact
For questions or issues regarding the dataset, please contact [Amazon Reviews 2023](https://amazon-reviews-2023.github.io/).