What’s Out There
← Back to swiping
Consumer & commercialFreeAcademic

Amazon Reviews 2023

McAuley Lab (UC San Diego)

571.5 million Amazon product reviews, 245 percent more than the previous version, through September 2023.

An updated academic release of Amazon reviews with ratings, review text, helpful votes, and rich item metadata including descriptions, categories, price, brand, and images. One row is one review, and a separate table holds item metadata. Timestamps are precise to the millisecond.

Access

Cost
FreeNo charge.
License
License not statedNot tagged in the Hub record; research dataset from Amazon data, check the page for terms
Sign-up
No sign-up
How you get it
API and Bulk downloadJSON Lines, Parquet

What's in it

One row is
One row per review, with a separate row per product in the item metadata table
Covers
Amazon marketplace (primarily US)
Years
May 1996 to September 2023
Updated
One-time release
Size
571.54 million reviews
ratingreview title and texttimestamphelpful votesverified purchaseproduct IDpricecategory

Quality

4 of 5

The standard large academic review dataset with benchmark splits for recommendation, but sourced from scraped commercial data of unclear terms.

  • Terms of use not stated on the record
  • Review text may include fake or incentivized reviews
  • Reviewer identities are pseudonymous IDs but still sensitive

Connects to

  • Amazon Reviews 2018: Previous version, which this release extends with newer reviews and cleaner metadata
  • IMDB Reviews: Smaller sentiment dataset for movie reviews

Questions it could answer

  • How do review ratings shift over time by category?
  • Which items have the largest gap between ratings and review text sentiment?

Similar data sets

  • UCSD Goodreads Datasets

    228 million shelvings, ratings, and reviews from 876,000 Goodreads readers, scraped in 2017.

    Film, TV, & booksUniversity of California San Diego (McAuley Lab)Free

  • MovieLens

    32 million movie ratings from 200,000 people: the benchmark behind decades of recommender research.

    Film, TV, & booksGroupLens Research, University of MinnesotaFree

  • Pushshift Reddit Archive

    The archive of nearly every Reddit post and comment, until Reddit shut researchers out in 2023.

    Social mediaPushshift (Network Contagion Research Institute)By application

  • IMDb Non-Commercial Datasets

    IMDb's core tables for millions of titles and people, plus ratings, refreshed daily as free TSVs.

    Film, TV, & booksIMDb (Amazon)Free

  • Review of Particle Physics (Particle Data Group)

    The physics bible of every known particle's mass, lifetime, and decay modes, updated since 1957.

    Physics & chemistryParticle Data Group, Lawrence Berkeley National Laboratory and international collaboratorsFree

  • Meta Content Library

    Public Facebook, Instagram, Threads, and WhatsApp Channels posts, for vetted researchers only.

    Social mediaMeta PlatformsBy application