clarin-knext/CLARIN-Emo

该数据集由波兰语的消费者评论组成，涵盖四个领域：酒店、医药、产品和大学。这些评论包括非观点的信息性文本，通常为中性。每个句子以及整个评论都标注了Plutchnik情绪轮中的情绪（如喜悦、信任、期待、惊讶、恐惧、悲伤、厌恶、愤怒）以及感知的情感（积极、消极、中性）。数据集由六人独立标注，最终标签由至少两人同意确定，因此可能存在对立情绪的标注。数据集分为训练集、验证集和测试集，分别包含776、167和167篇评论。

The dataset is made up of consumer reviews written in Polish. Those reviews belong to four domains: hotels, medicine, products, and university. This collection also contains non-opinion informative texts belonging to the same domains (meaning they are mostly neutral). Each sentence, as well as all the reviews as a whole, are annotated with emotions from the Plutchniks wheel of emotions (joy, trust, anticipation, surprise, fear, sadness, disgust, anger), as well as the perceived sentiment (positive, negative, neutral), with ambivalent sentiment being labeled using both positive and negative labels. The dataset was annotated by six people who did not see each others decisions. These annotations were aggregated by selecting labels annotated by at least 2 out of 6 people, meaning controversial texts and sentences can be annotated with opposing emotions. The dataset is split into training, validation, and test sets, containing 776, 167, and 167 reviews respectively.