Heading & Section Structure
The document's organization — what chunking, section-scoped retrieval, citation and navigation run on; fails independently of order
What this capability means
A tool has this capability when it preserves the document’s heading hierarchy and section structure so the output still reflects how the source is organised for chunking, section-scoped retrieval, citation, and navigation. Headings should remain real structure, with their nesting intact, rather than being flattened into plain paragraphs or losing section boundaries.
Boundary: It does not judge whether text is read in the right order across the page; that belongs to Reading Order & Layout.
Scenarios that test this capability
A scenario is a real-world situation used to test a capability.
Benchmarks that include this capability
A capability has global identity and may be used by more than one benchmark.
Converting a complex PDF into clean Markdown with an open-source library
Also evaluates this capability.
Converting a complex PDF into clean Markdown with a hosted API
Which hosted API converts a complex, real-world PDF into faithful, usable Markdown?