This dataset contains OCR image–markdown pairs specifically curated for document structure retrieval and reconstruction tasks,focusing on Arabic and English documents across a variety of domains.

Dataset Summary

Source of Information:This dataset was created by crawling publicly accessible Arabic and English PDFs from:

Official Arabic government document portals Arabic news websites and online magazines Community and… See the full description on the dataset page: https://huggingface.co/datasets/presightai/arabic_doc_to_markdown.