agentlans/Universal-NER-Pile-NER

该数据集仓库提供了经过筛选的Universal-NER数据集子集，这些子集被重新格式化以便于分析和集成到下游任务中，如特征提取和标记分类。数据集按命名实体提取等任务类别进行分类。仓库包括各种配置的数据文件，每个文件包含不同大小的子集，以及两种用于通用实体识别的定义和类型的具体配置。数据字段包括唯一标识符、输入文本、检测到的和未检测到的实体、实体的JSON映射以及数据的来源。该数据集遵循知识共享署名非商业性4.0国际许可（CC BY-NC 4.0）。

This dataset repository provides curated subsets of the Universal-NER dataset, formatted for easier analysis and integration into downstream tasks such as feature-extraction and token-classification. The datasets are categorized under tasks like named entity extraction. The repository includes various configurations of data files, each with a different subset size, and two specific configurations for universal entity recognition with definitions and types. The data fields include unique identifiers, input text, detected and undetected entities, a JSON mapping of entities, and the source of the data. The dataset is licensed under Creative Commons Attribution Non-Commercial 4.0 International (CC BY-NC 4.0).