DAMO-NLP-SG/SOUL
收藏Hugging Face2023-10-16 更新2024-03-04 收录
下载链接:
https://hf-mirror.com/datasets/DAMO-NLP-SG/SOUL
下载链接
链接失效反馈官方服务:
资源简介:
---
license: mit
task_categories:
- text-classification
- text-generation
language:
- en
size_categories:
- 10K<n<100K
---
This repo contains the data for our paper "SOUL: Towards Sentiment and Opinion Understanding of Language" in EMNLP 2023.
[Github repo](https://github.com/DAMO-NLP-SG/SOUL/tree/main)
## Statistics
The SOUL dataset comprises 15,028 statements related to 3,638 reviews, resulting in an average of 4.13 statements per review. To create training, development, and test sets, we split the reviews in a ratio of 6:1:3, respectively.
| Split | # reviews | # statements | True | False | Not-given | Total |
|-------|-----------|--------------|------|-------|-----------|-------|
| Train | 2,182 | 3,675 | 2,159| 8,834 | 3,000 | 8,834 |
| Dev | 365 | 617 | 503 | 1,481 | 361 | 1,481 |
| Test | 1,091 | 1,956 | 1,664| 4,713 | 1,093 | 4,713 |
| Total | 3,638 | 6,248 | 3,613|15,028 | 5,167 |15,028 |
提供机构:
DAMO-NLP-SG
原始信息汇总
数据集概述
许可证
- MIT许可证
任务类别
- 文本分类
- 文本生成
语言
- 英语
数据集大小
- 10K<n<100K
数据集统计
- 数据集名称: SOUL
- 数据来源: EMNLP 2023论文"SOUL: Towards Sentiment and Opinion Understanding of Language"
- 数据集组成: 包含15,028条陈述,涉及3,638条评论,平均每条评论有4.13条陈述。
- 数据集划分:
- 训练集: 2,182条评论,3,675条陈述
- 开发集: 365条评论,617条陈述
- 测试集: 1,091条评论,1,956条陈述
- 总计: 3,638条评论,6,248条陈述



