UCSD Goodreads Datasets
University of California San Diego (McAuley Lab)
228 million shelvings, ratings, and reviews from 876,000 Goodreads readers, scraped in 2017.
Academic data sets built from public Goodreads shelves in late 2017: metadata for 2.36 million books (works, series, authors), 228.6 million user-book interactions (reads and ratings), and about 15 million reviews, with genre subsets and an English spoiler-annotated subset. Each interaction row is one user's shelving, rating, or review of a book. It is the largest public book-reading data set.
Access
- Cost
- FreeFree download; academic use only.
- License
- Open, non-commercial onlyAcademic use only; no redistribution or commercial use
- Sign-up
- No sign-up
- How you get it
- Bulk downloadJSON (gzip)
What's in it
- One row is
- One row per user-book interaction; one record per book or review
- Covers
- Goodreads users worldwide (mostly English-language)
- Years
- Activity through late 2017
- Updated
- One-time release
- Size
- 2.36 million books; 876,145 users; 228.6 million interactions; about 15 million reviews
Quality
Very large and documented in published papers, but frozen in 2017 and drawn only from public shelves.
- No updates since 2017
- Goodreads users skew toward avid, English-language readers
- Files up to about 11 GB
Connects to
- Open Library Data Dumps: Match on ISBN for edition and subject metadata
- Seattle Public Library Checkouts by Title: Compare what readers rate with what they borrow
- MovieLens: Parallel ratings benchmark for films
Questions it could answer
- Do series lose readers book by book, and how steeply?
- Are long books rated higher by those who finish them?