gtfintechlab/finer-ord
收藏资源简介:
FiNER-开放研究数据集(FiNER-ORD)包含从webz.io收集的金融新闻文章的手动标注数据集(英文)。总共有47851篇新闻文章,每篇文章以JSON文档形式提供,包含文章来源、发布日期、作者和标题等元数据信息。从所有文章中随机抽取220篇进行手动标注,过滤掉空文章后,最终得到201篇文章。使用Doccano开源标注工具对人物(PER)、地点(LOC)和组织(ORG)实体进行手动标注。数据集分为训练集、验证集和测试集,标注过程由不同的标注者完成,以确保无偏结果。标注后的文章被分割成句子,并对每个句子进行分词处理,多词实体被分为多个标记(如PER_B表示人物实体的开始标记,PER_I表示中间标记)。数据集的详细统计数据可在数据分割部分查看。
FiNER-Open Research Dataset (FiNER-ORD) is an English manually annotated financial news article dataset collected from webz.io. In total, 47,851 news articles are included, each presented as a JSON document with metadata such as article source, publication date, author, and title. A random sample of 220 articles was selected from the full corpus for manual annotation, and 201 valid articles were finally retained after filtering out empty articles. The open-source annotation tool Doccano was utilized to manually annotate three entity types: Person (PER), Location (LOC), and Organization (ORG). The dataset is split into training, validation, and test subsets. To ensure unbiased results, the annotation work was completed by multiple independent annotators. Annotated articles are first segmented into sentences, and each sentence is then tokenized. Multi-word entities are split into multiple tokens (e.g., PER_B represents the beginning token of a person entity, while PER_I represents the inside token of the same entity). Detailed statistical information of the dataset can be found in the data split section.
数据集概述
数据集名称
- 名称: FiNER-Open Research Dataset (FiNER-ORD)
数据集详情
- 语言: 英语
- 类别: 单语种
- 大小: 1K<n<10K
- 任务: 命名实体识别
- 许可证: cc-by-nc-4.0
数据集内容
- 来源: 金融新闻文章,从webz.io收集
- 数量: 总共47851篇新闻文章,其中201篇用于手动标注
- 格式: JSON文档,包含文章来源、发布日期、作者和标题等元数据
- 标注: 使用Doccano工具手动标注人名(PER)、地点(LOC)和组织(ORG)实体
数据集结构
- 训练集大小: 1.08 MB
- 验证集大小: 135 KB
- 测试集大小: 336 KB
- 数据字段:
doc_idx,sent_idx,gold_token,gold_label - 标签集: {O: 0, PER_B: 1, PER_I: 2, LOC_B: 3, LOC_I: 4, ORG_B: 5, ORG_I: 6}
数据分割统计
| FiNER-ORD | Train | Validation | Test |
|---|---|---|---|
| # Articles | 135 | 24 | 42 |
| # Tokens | 80,531 | 10,233 | 25,957 |
| # LOC entities | 1,255 | 267 | 428 |
| # ORG entities | 3,440 | 524 | 933 |
| # PER entities | 1,374 | 222 | 466 |
联系方式
- 联系人: Agam Shah, Ruchit Vithani
- 邮箱: ashah482[at]gatech[dot]edu, rvithani6[at]gatech[dot]edu
- GitHub: @shahagam4, @ruchit2801
- 网站: https://shahagam4.github.io/




