IMDB Large Movie Review Dataset is a dataset suitable for binary classification of sentiment. It is designed to serve as a benchmark for sentiment classification. The dataset contains 50,000 polarized movie reviews for training and testing, and 50,000 unlabeled data are available. This dataset was released by Stanford University in 2011, and the related paper is "Learning Word Vectors for Sentiment Analysis".