Deep Document Understanding

Deep Document Understanding — your content is read, not just indexed

Other searches find files. RAGflow understands their content — even in scans, images, and tables.

The RAGflow Moodle Suite answers questions and searches content not based on filenames and full text, but by reading the actual content of your documents. This is exactly what sets it apart from the classic Moodle search.

Scans, images, and tables too

OCR — scanned PDFs and photographed pages become searchable text (multilingual).
Layout understanding — titles, paragraphs, tables, figures, and formulas are recognized as structure and read in the correct order.
Image understanding — a vision model describes diagrams and screenshots; the description becomes searchable, and the image appears as a citation — in chat answers and in the search.

Meaning instead of keywords

Questions and documents end up in the same vector space — what is found is what means the same thing, even if phrased differently. An optional rerank model sharpens the order of results.

Selectable parsers

In addition to the built-in DeepDoc, MinerU and Docling (RAGflow 0.25+) are available with formula and table recognition — selectable per knowledge base.

The difference from Moodle search

	Standard Moodle search	RAGflow (this suite)
Searches in	Titles, text fields, metadata, forum posts	the content of the files themselves
Scanned PDFs / images	invisible	OCR + image description → searchable
Tables / diagrams	ignored	captured as structure or via vision model
Matches by	exact keyword	meaning (embeddings, optional rerank)
Result	Link to the activity	the supporting passage + figure, optionally as an answer

Note: OCR, layout, and table recognition are built-in. Understanding image content additionally requires a vision model (img2txt) in RAGflow.

Learn more in the documentation: docs.ragcon.ai/moodle-ragflow/document-understanding.

Try RAGflow

RAGflow runs entirely on European infrastructure and adapts flexibly to your requirements—from entry level to enterprise solution.