Hotter and Colder
收藏资源简介:
Hotter and Colder是由冰岛大学等机构创建的一个数据集,旨在分析冰岛博客评论中的各种在线行为。该数据集基于GPT-4o mini模型对约800,000条博客评论进行自动标注,并经过人工复核,最终包含12,232条独立注释和19,301条标注。数据集涵盖了情感分析、情绪检测、仇恨言论、群体概括等25个任务,为研究内容审查和自动检测有害在线行为提供了一个重要的资源。
Hotter and Colder is a dataset created by institutions including the University of Iceland, designed to analyze various online behaviors in Icelandic blog comments. This dataset automatically annotates approximately 800,000 blog comments using the GPT-4o mini model, followed by manual review, and ultimately contains 12,232 independent annotations and 19,301 labeled entries. The dataset covers 25 tasks including sentiment analysis, emotion detection, hate speech detection, group generalization, and others, providing a crucial resource for research on content moderation and automated detection of harmful online behaviors.




