deeplang-ai/StructBench
收藏Hugging Face2025-12-18 更新2025-12-20 收录
下载链接：
https://hf-mirror.com/datasets/deeplang-ai/StructBench
下载链接
链接失效反馈官方服务：
资源简介：
StructBench是一个用于评估细粒度文档结构分析的基准测试集。它提供了一个高质量的测试集，包含248个不同格式的文档，其中203个是网页，47个是PDF。为确保可靠的基准真值，所有文档都经过了解析和句子分割，并由人类专家手动标注了语篇结构。除了结构化标注外，还包括了原始的网页和PDF文件。
StructBench is a benchmark for evaluating fine-grained document structure analysis. It provides a high-quality test set of 248 documents in diverse formats, including 203 Web pages and 47 PDFs. To ensure reliable ground truth, all documents were: Parsed and sentence-segmented, Manually annotated by human experts for discourse structure. In addition to the structured annotations, raw Web pages and PDF files are included.
提供机构：
deeplang-ai
