KoCHET
收藏KoCHET: 韩国文化遗产语料库
数据集概述
KoCHET是一个用于实体相关任务的韩国文化遗产语料库,包括命名实体识别(NER)、关系抽取(RE)和实体类型(ET)。
下载链接
数据集下载链接:KoCHET
文件夹结构
KoCHET目录结构:- 命名实体识别(NER):
ner_{train,dev,test}.jsoncoarse_list.txt
- 关系抽取(RE):
re_{train,dev,test}.jsonrelation_list.txt
- 实体类型(ET):
et_{train,dev,test}.jsonfine_grained_list.txt
- 命名实体识别(NER):
数据集统计
| 任务 | 指标 | 训练集 | 开发集 | 测试集 |
|---|---|---|---|---|
| NER | 示例数量 | 89,884 | 11,245 | 11,233 |
| 实体数量 | 393,076 | 32,003 | 32,153 | |
| RE | 示例数量 | 31,012 | 3,876 | 3,877 |
| 关系数量 | 64,080 | 8,031 | 7,831 | |
| ET | 示例数量 | 90,558 | 11,320 | 11,320 |
| 提及数量 | 266,209 | 33,226 | 33,395 |
基线模型性能
| 模型 | NER实体F1宏 | NER实体F1微 | NER字符F1宏 | NER字符F1微 | RE F1宏 | RE F1微 | ET F1宏 | ET F1微 |
|---|---|---|---|---|---|---|---|---|
| mBERT | 59.81 | 58.99 | 71.80 | 90.44 | 80.85 | 89.94 | 91.64 | 91.60 |
| XLM-RoBERTa-base | 76.57 | 79.67 | 82.69 | 94.77 | 80.29 | 89.79 | 91.13 | 91.00 |
| KLUE-BERT-base | 39.31 | 33.40 | 55.63 | 81.06 | 82.44 | 90.77 | 93.08 | 93.02 |
| KLUE-RoBERTa-base | 38.92 | 33.48 | 55.47 | 81.08 | 82.42 | 90.78 | 92.80 | 92.84 |
引用
bibtex @inproceedings{kim-etal-2022-kochet, title = "{K}o{CHET}: A {K}orean Cultural Heritage Corpus for Entity-related Tasks", author = "Kim, Gyeongmin and Kim, Jinsung and Son, Junyoung and Lim, Heuiseok", booktitle = "Proceedings of the 29th International Conference on Computational Linguistics", month = oct, year = "2022", address = "Gyeongju, Republic of Korea", publisher = "International Committee on Computational Linguistics", url = "https://aclanthology.org/2022.coling-1.308", pages = "3496--3505", abstract = "As digitized traditional cultural heritage documents have rapidly increased, resulting in an increased need for preservation and management, practical recognition of entities and typification of their classes has become essential. To achieve this, we propose KoCHET - a Korean cultural heritage corpus for the typical entity-related tasks, i.e., named entity recognition (NER), relation extraction (RE), and entity typing (ET). Advised by cultural heritage experts based on the data construction guidelines of government-affiliated organizations, KoCHET consists of respectively 112,362, 38,765, 113,198 examples for NER, RE, and ET tasks, covering all entity types related to Korean cultural heritage. Moreover, unlike the existing public corpora, modified redistribution can be allowed both domestic and foreign researchers. Our experimental results make the practical usability of KoCHET more valuable in terms of cultural heritage. We also provide practical insights of KoCHET in terms of statistical and linguistic analysis. Our corpus is freely available at https://github.com/Gyeongmin47/KoCHET.", }




