Amazon Polarity

该数据集是针对亚马逊商品评论的二元分类任务设计的，旨在判断评论是积极的（四星或五星）还是消极的（一星或二星）。此外，该数据集也用于评估取消学习方法的性能。该数据集包含了25,000个样本（经过子采样处理），其任务是对评论进行情感分析。

This dataset is designed for binary classification tasks on Amazon product reviews, with the objective of classifying a review as positive (4 or 5 stars) or negative (1 or 2 stars). Furthermore, it is also utilized to evaluate the performance of unlearning methods. The dataset contains 25,000 subsampled samples, and its associated task is sentiment analysis for product reviews.