CSKB Population Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/HKUST-KnowComp/PseudoReasoner
下载链接
链接失效反馈官方服务：
资源简介：
该数据集旨在训练和评估常识知识库（CSKB）填充模型，它与KG-BERT和COMET等模型结合使用，以执行常识知识库填充任务。该数据集包含一百万个标注样本用于训练，以及两亿个候选空间用于填充任务。其核心任务是常识知识库的填充。
This dataset is designed for training and evaluating commonsense knowledge base (CSKB) completion models. It is used in conjunction with models such as KG-BERT and COMET to perform commonsense knowledge base completion tasks. The dataset contains 1 million labeled training samples and 200 million candidate entries for the completion task. Its core task is commonsense knowledge base completion.
提供机构：
HKUST-KnowComp
