Source Ingestion and PDF Fidelity¶
Use this workflow when a research node, manuscript claim, or teaching chapter depends on a math-heavy paper. Zotero is the library of record for metadata, collections, and CRUD operations. Local PDF bundles remain necessary for theorem-fidelity checks because extracted text alone often corrupts equations, theorem numbering, and notation.
Source Priority¶
- Use arXiv HTML or TeX source when available.
- Use publisher HTML when it preserves theorem and equation structure.
- Use a Zotero-backed local PDF bundle with page text plus page images.
- Use raw PDF text only for search, never as the sole source for notation-sensitive claims.
Zotero Role¶
Use Zotero for:
- collection membership and de-duplication,
- citation metadata and BibTeX snapshots,
- attachment discovery,
- item creation, updates, tags, notes, and deletions when explicitly requested.
The project-wide bibliography snapshot is ../10_conjectures.bib. Refresh or check Zotero before assuming a paper is absent.
Local PDF Bundle Role¶
Use local bundles for:
- theorem and lemma statement verification,
- page-level citation audits,
- equation-sensitive reading,
- manuscript source-fidelity checks.
Durable project bundles live under ../sources/paper-bundles/. A bundle should contain manifest.json, text.txt, per-page pages/page-XXXX.txt, and page images when available.
Commands¶
Resolve local Zotero attachments from the bibliography snapshot:
Create or refresh a paper bundle:
Node-Writing Rule¶
When adding or revising a research node, classify every important claim as one of:
- established theorem,
- source-backed definition,
- inferred bridge,
- open gap,
- boundary evidence.
If a formula or theorem number matters, verify it against page images or source-formatted text before treating it as established.