GoodNewsEveryone
收藏资源简介:
GoodNewsEveryone数据集由斯图加特大学机器语言处理研究所创建,包含5000条来自82个不同来源的英文新闻标题。数据集通过众包方式进行了情感类别、情感强度、语义角色(体验者、原因、目标、线索)以及读者视角的标注。该数据集旨在解决情感分析中的细粒度分析问题,支持情感分类、情感强度预测、情感原因检测等研究,并提供了一个两阶段的标注程序和基准模型结果。
The GoodNewsEveryone dataset was created by the Institute for Machine Language Processing at the University of Stuttgart, containing 5,000 English news headlines from 82 distinct sources. The dataset was annotated via crowdsourcing for sentiment categories, sentiment intensity, semantic roles (experiencer, cause, target, cue), and reader perspective. It aims to address fine-grained analysis problems in sentiment analysis, supporting research such as sentiment classification, sentiment intensity prediction, and sentiment cause detection, and provides a two-stage annotation procedure and benchmark model results.

- 1GoodNewsEveryone: A Corpus of News Headlines Annotated with Emotions, Semantic Roles, and Reader Perception斯图加特大学机器语言处理研究所 · 2020年



