# TSMPD-US Public Parquet Embeddings

This directory contains SBERT-generated vector embeddings for the TSMPD-US-Public v1.0 dataset.

## What's Included
- Chunked `.parquet` files containing:
  - `uid`: unique product identifier
  - `vendor`: merchant name
  - `title`: product title
  - `paragraph`: product description
  - `embedding`: list of 384-dimensional float values (SBERT embedding)

## Use Cases
- **Semantic product search** using dense vector similarity
- **LLM grounding** to enrich responses with small business products
- **RAG (Retrieval-Augmented Generation)** pipelines
- **Product classification and clustering** based on description similarity

## Licensing
This data is licensed under the [Open Data Commons Attribution License v1.0 (ODC-By 1.0)](https://opendatacommons.org/licenses/by/1.0/).
SBERT embeddings are generated using the `sentence-transformers/all-MiniLM-L6-v2` model, which is licensed under Apache 2.0.

## Notes
These embeddings correspond to the text-only public dataset (no images, URLs, or variant data).