IMDB movie reviews
50,000 movie reviews labelled positive or negative - a binary text-classification benchmark balanced by construction.
A sentiment-classification set of 50,000 reviews split evenly between training and test, and evenly between positive and negative labels, so a classifier that always guesses one class scores exactly 50%. Chollet uses it as the introductory text-classification example. Because the input is raw prose, it is also the natural place to see why text has to be tokenised and embedded before a network can read it.
Key columns
Representative fields, refer to the source for the full, authoritative schema.
| Column | Type | Description |
|---|---|---|
| review | string | The raw review text, unprocessed. |
| sentiment | string | Either "positive" or "negative"; the two classes are balanced. |
License: Released for research use by the authors; see the source page for terms.