frasem/sentiment-dataset

TweetEval数据集是一个包含七种异构任务的Twitter多类推文分类基准测试集，当前配置仅展示情感分析任务。所有数据集均以相同格式和固定的训练、验证及测试分割呈现。数据集中的文本为Twitter用户使用的英语，情感标签分为负面（0）、中性（1）和正面（2）。数据集由Cardiff NLP团队创建，需遵守Twitter的服务条款和API使用条款。

TweetEval consists of seven heterogenous tasks in Twitter, all framed as multi-class tweet classification. This configuration exposes the sentiment task only. All datasets have been unified into the same benchmark, with each dataset presented in the same format and with fixed training, validation and test splits. The text in the dataset is in English, as spoken by Twitter users, with labels for negative (0), neutral (1), and positive (2) sentiments. The dataset was created by the Cardiff NLP team and requires compliance with Twitters Terms of Service and API Terms of Service.