Structured Markdown content of all monocorpus documents that have parsed content (content_url), packaged in the same format as yasalma/tt-structured-content.

57,730 documents, ~3.25B characters.
57,238 documents also exist in tt-structured-content (same id); their text here is the current content from monocorpus, which in most cases differs from (is newer than) the tt-structured-content copy.
285 documents are not in tt-structured-content (pptx/html/epub/txt/docx/fb2 etc.).
207 documents (tatar_structured_content_0007.parquet) are copied unchanged (id, publish_year, genre, text) from tt-structured-content because monocorpus has no parsed content for them (141 are not in monocorpus at all). 41 are very short (<200 chars, mostly "Скачано с www.znanio.ru" stubs).

Dataset Structure

Same columns as tt-structured-content:

id: monocorpus document md5
publish_year: publication year (from monocorpus; falls back to tt-structured-content), may be missing
genre: comma-separated genres. Taken from tt-structured-content (Gemini-inferred) by id == md5; for documents not in tt-structured-content, taken from the monocorpus schema.org metadata genre list
text: document content in Markdown

Each row is one full document.

Notes

7 monocorpus documents whose content archive contains an empty Markdown file are excluded.
Includes all documents with parsed content, regardless of monocorpus sharing_restricted flag.

Downloads last month: -

Total file size:

1.44 GB