vmoiseev20/UAReviews

UAReviews是一个包含11,580条乌克兰语用户评论和反馈的基准数据集，标注了情感和意图类别。该数据集旨在评估和微调乌克兰语的情感、情绪和意图理解模型。数据来源包括乌克兰数字化转型部提供的原始评论和从COSMUS数据集中选取并重新标注的子集，以平衡低频率类别。数据集结构包含唯一标识符、原始文本、评分、来源、最终情感、最终类别、分割和文本长度等字段。数据集采用CC BY 4.0许可。

UAReviews is a curated benchmark of 11,580 Ukrainian user reviews and feedback comments labeled for both emotion and intent category. It is designed for evaluating and fine-tuning sentiment, emotion, and intent-understanding models for the Ukrainian language. The dataset includes original user reviews from the Ministry of Digital Transformation of Ukraine and a re-labeled subset from the COSMUS dataset to balance underrepresented categories. The dataset structure includes fields such as unique identifier, original text, rating, source, final emotion, final category, split, and length. The dataset is licensed under CC BY 4.0.