IMDb: Large Movie Review Dataset for Binary Sentiment Classification

A set of 50,000 highly polar movie reviews for binary sentiment classification, with 25,000 reviews designated for training and 25,000 for testing. The dataset, created by Drane111, also includes additional unlabeled data for use. It was last updated on the platform in 2026.

Use Cases

• Train binary sentiment classifiers based on the labeled movie reviews.
• Benchmark NLP model performance on a large-scale text classification task.
• Use the unlabeled data for semi-supervised learning or self-training experiments.
• Analyze language patterns in polar movie reviews for linguistic research.

Strengths

• Contains 50,000 total labeled reviews, which is a substantial scale for model training.
• Provides a clear split of 25,000 reviews for training and 25,000 for testing.
• Includes additional unlabeled data, which can be used for expanded experimentation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.
• The last updated date is 2026-05-31, which may indicate a future timestamp requiring verification.

Provenance

Source

Drane111 on Hugging Face.

Freshness

Last updated 2026-05-31 06:31:06; freshness should be verified.