BEE-spoke-data/govdocs1-pdf-source

govdocs1数据集是一个包含约220,000个开放访问PDF文档的数据集，总页数约为6.6M页，主要来源于政府文件。这些文档是为了进行OCR（光学字符识别）而提取的。数据集以`tar`文件格式上传，并分为多个部分以适应大小和文件数量限制。此外，还有一个包含5,000个随机抽样PDF文档的样本集，这些可以在浏览器中直接预览。数据集的元数据存在一定程度的缺失，尤其是主题字段缺失率高达79%。

The govdocs1 dataset is a collection of approximately 220,000 open-access PDF documents, totaling about 6.6M pages, primarily sourced from government files. These documents are extracted for OCR (Optical Character Recognition) processing. The dataset is uploaded in the `tar` file format and split into multiple parts due to size and file count limits. Additionally, there is a sample set containing 5,000 randomly sampled PDF documents that can be previewed directly in the browser. The dataset has some metadata completeness issues, particularly with the Subject field having a missing rate as high as 79%.