iahlt/arabic_ner_mafat

这是一个阿拉伯语命名实体识别数据集，提供了文本的跨度级别和标记级别标注。它包含粗粒度标注（如MISC，ORG，LOC等），出现在spans字段中，以及以BILUO格式表示的细粒度标注，出现在raw_tags字段中，以及对应的数字ID在ner_tags字段中。数据集还包含了关于文档的额外元数据信息。

This is an Arabic Named Entity Recognition dataset that provides spans-level and token-level annotations for Arabic text. It includes coarse-grained annotations (such as MISC, ORG, LOC, etc.) appearing in the spans field, fine-grained annotations in BILUO format (e.g., B-MISC, I-MISC, L-MISC) in the raw_tags field, and their corresponding numeric IDs in the ner_tags field. The dataset also contains additional metadata information about the document.