All datasets / Search & relevance

Amazon ESCI

Query-to-product relevance judgments for evaluating ecommerce search and substitute recommendations.

View source on GitHub ↗Review required

What the resource contains

  • Queries with IDs and locales
  • Product titles, descriptions, brands and attributes
  • Exact, Substitute, Complement and Irrelevant labels; benchmark splits

Potential uses

  • Train search rerankers
  • Classify search relevance
  • Evaluate substitute-product suggestions

Access & formats

Parquet. Repository files. Follow the project’s instructions on GitHub; some resources require external files, registration or approval.

License & reuse

The repository declares Apache-2.0. Confirm its scope and any upstream product-content terms before commercial reuse.

Important limitations

Labeled benchmark data, not a current inventory or a record of sales. Preserve the supplied evaluation splits.

Source & attribution

amazon-science — original GitHub project ↗

Documentation reviewed 2026-09-24. Review scope and licensing notes. Suggest a correction.

Related resources