Yelp Sentiment

该数据集的每个样本都包含了一个性别标注及其相应的置信度值。此外，每个样本的情感标注是通过在5800万条推文上训练的RoBERTa模型完成的。对于那些性别标注和情感标注的置信度都大于0.85的样本，我们将其保留用于情感分析任务。

Each sample in this dataset includes a gender annotation and its corresponding confidence score. Additionally, the sentiment annotation for each sample is generated using a RoBERTa model trained on 58 million Tweets. For samples where both the confidence scores of the gender annotation and the sentiment annotation exceed 0.85, we retain them for the sentiment analysis task.