pd3f-dataset-bmjv

Dataset of (mostly German) PDFs used to develop pd3f.

This repository contains the code to scrape and download some public documents (PDFs). The can files be downloaded here: https://data.jfilter.de/nlp/pd3f/bmjv_v1.zip.

Origin of the Dataset

Downloaded "Stellungnahmen zu Referententwürfen" from the BMJV, around 02.04.2022
Prepend filenames with numbers
OCRd for German and English with OCRmyPDF
Sort / group by language
Redo broken OCR (manually detecting errors while working on the PDFs)

License

GPLv3