rams-event-dataset
收藏资源简介:
该数据集为ACE(自动内容抽取)事件检测数据集的元数据文件集合,包含用于持续学习的任务流配置、事件类型标签与数值ID的映射关系、事件类型的频率分布统计以及ID与token表示的映射关系。具体包含四个JSON文件:streams.json(5个配置项)、label2id.json(139个事件类型标签)、label_freq.json(事件类型频率分布)和id2tokens.json(140个ID映射)。这些元数据文件主要用于支持基于ACE数据集的事件检测任务中的持续学习场景,但未包含原始ACE数据集的具体内容或样本数据。使用方式可通过Hugging Face Hub下载并加载相关配置文件。
This dataset is a collection of metadata files for the ACE (Automatic Content Extraction) event detection dataset. It encompasses task flow configurations for continual learning, mappings between event type labels and numeric IDs, frequency distribution statistics of event types, and mappings between IDs and token representations. Specifically, it contains four JSON files: "streams.json" (with 5 configuration items), "label2id.json" (covering 139 event type labels), "label_freq.json" (storing event type frequency distributions), and "id2tokens.json" (containing 140 ID mappings). These metadata files are primarily used to support continual learning scenarios in event detection tasks based on the ACE dataset, but do not include the specific content or sample data of the original ACE dataset. Users can download and load the relevant configuration files via the Hugging Face Hub for application.



