FactCheck-AI/FactCheck

FactCheck是一个用于评估大型语言模型在知识图谱事实验证方面的基准数据集。它整合了来自YAGO、DBpedia和FactBench的结构化事实，以及通过网络搜索获取的证据，包括问题、摘要、全文和元数据等。数据集旨在用于句子级别的事实检查和问答任务，并提供了一个训练集分割。每个示例都包含丰富的元数据字段，如唯一ID、数据源、问题、排名、URL、标题、摘要、文本、媒体资源等。

FactCheck is a benchmark dataset for evaluating large language models on knowledge graph fact verification. It combines structured facts from YAGO, DBpedia, and FactBench with web-extracted evidence including questions, summaries, full text, and metadata. The dataset is designed for sentence-level fact-checking and QA tasks, and provides a train split. Each example includes rich metadata fields such as unique ID, source dataset, question, rank, URL, title, summary, text, media assets, etc.