BramVanroy/conll2003

CoNLL-2003数据集是一个用于语言无关的命名实体识别的数据集，包含对人物、地点、组织和杂项实体的标注。数据集为英语单语种，标注工作由众包完成。数据集分为训练集、验证集和测试集，每个实例包含单词、词性标签、分块标签和命名实体标签。数据集基于路透社语料库，并受特定的版权协议约束。数据集与seqeval指标兼容，用于评估。

The CoNLL-2003 dataset is for language-independent named entity recognition, including annotations for persons, locations, organizations, and miscellaneous entities. It is monolingual in English and has been crowdsourced. The dataset is split into training, validation, and test sets, with each instance containing words, part-of-speech tags, chunk tags, and named entity tags. Based on the Reuters Corpus, the dataset is licensed under specific agreements with Reuters Ltd and/or Thomson Reuters. It is compatible with the seqeval metric for evaluation.