ParsTwiner
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/overfit-ir/parstwiner
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为ParstwiNER，专为非正式实体识别而设计，它作为一个基准，用于评估在标准正式和非正式环境中的适应性。该数据集基于CONLL格式，包含21种实体类型，并且以其极低噪音的特点而著称。其任务是对命名实体的识别。
This dataset, named ParstwiNER, is specifically developed for informal named entity recognition (NER). It serves as a benchmark for evaluating model adaptability across standard formal and informal scenarios. Built on the CONLL format, this dataset covers 21 entity types and is renowned for its extremely low noise level. The core task of this dataset is named entity recognition.
