boluopai.Sources and verification
Choose sources for the missing evidence; these are starting points, not a mandatory queue or guarantees of availability.
| Source | Contribution | Verification limit |
|---|---|---|
| Author CV/site, publisher/journal | Identity, authoritative bibliography and publication relation | Reconcile dates, affiliations and versions against printed text. |
| Crossref, OpenAlex | DOI/title/author metadata and relationships | A record or closed-access flag does not exhaust full-text discovery. |
| Scholar, PhilArchive, PhilSci-Archive, university repositories | Versions and disciplinary manuscripts | Follow original record; verify file, author, deposit version and completeness. |
| CORE, BASE, Unpaywall | Repository/open-location leads | APIs may require registration, email or keys; follow current official docs. Empty results are bounded lookup results. |
| CNKI, NSSD, Chinese journal/university sites | Chinese bibliography and available text | Login, entitlement and downloadable format require separate evidence. |
| DOAB, OAPEN, library catalogues | Open books, editions, holdings | Catalogue, preview, borrowable record and downloadable whole book differ. |
For books compare title/copyright pages, contents, chapter range, final pages and references with the publisher's edition. An ISBN identifies a particular edition/format, not every translation. A chapter DOI and whole-book DOI name different objects. Check ISBN checksums when reconciling identifiers; retain source attribution instead of inventing missing identifiers.
For local PDF checks install Poppler and use
pdfinfo, pdftotext, and optional pdftoppm -f 1 -singlefile -png candidate.pdf first.
The inspector's JSON is evidence for byte integrity and parsing, not an automated
bibliographic verdict. Image-only text extraction requires visual inspection or
explicitly labelled OCR. Page count must be compared with expected pagination;
missing internal pages can coexist with plausible endpoints. Keep SHA-256 of
acquired bytes separate from a catalogue-advertised MD5. Never identify a missing
complete work by the hash of an incomplete scan or HTML response.
This public adaptation preserves the private-origin method's distinctions between relevance, identity, versions, access and acquisition gaps. It replaces host-specific authenticated capture scripts and incident logs with standard HTTP/browser download and local PDF inspection. No private library-client configuration or third-party implementation is bundled. External tools and retrieved publications retain their own licenses and access conditions.