ChemRxiv source processing method

Authority

Use the collection scope to decide what to retain and how to represent it. The main paper and acquired supporting information are the scientific sources. Files and acquisition notes in data/YYYY-MM-DD/ establish only which sources were acquired or absent. Follow the established paper-page and dated-index formats and use the validator without gaming it. Existing pages are structural examples, not scientific sources or precedents for eligibility. The validator checks paper-page structure and notation, plus dated-index paper links and titles. It does not prove scientific accuracy, scope, source coverage, or the index’s analysis-availability claims.

Procedure

1. Establish the source set

   Inventory the main paper and every acquired supporting-information source for the same versioned DOI in data/YYYY-MM-DD/. Do not mix versions or use existing HTML as a scientific source.

   Inspect every acquired source completely with a safe, read-only method appropriate to its format. File format does not determine eligibility. If a source cannot be inspected completely, or an eligible record cannot be given a source locator accepted by the scope and validator, leave the paper incomplete and do not write its page or index row. Do not discard a source or treat partial coverage as complete. The local source set proves only what was acquired for this repository.

2. Inspect every source

   Use searchable text to locate candidate analytical records, then verify them in a directly inspected source representation. Inspect every PDF page and verify retained records against rendered pages. Read written results in captions, legends, and annotations; do not infer values, trends, peak lists, or assignments from the graphics. For other formats, inspect all content that can carry documentary information, including tables, sheets, notes, annotations, and embedded documents; recursively inspect documentary files inside archives. Never execute macros, scripts, embedded programs, or other active content. A search hit is not proof of a complete record, and a failed search is not proof that no record exists.

3. Write the paper page

   Apply the collection scope record by record, follow the established paper-page format, and satisfy the validator. Name the HTML after the main-paper source stem, without its extension, and place it in ext/YYYY-MM-DD/. Read the authors, title, year, and versioned DOI from the main paper; render the title in sentence case. In the paper bibliography, link only the DOI.

   Apply exclusions to individual results within mixed blocks. Keep the conditions and qualifications attached to eligible results, and preserve useful coverage notes as specified in the scope. A note that image data were not transcribed is not itself an analytical result.

   Create a negative page only after complete inspection of every acquired source finds no eligible analytical result; coverage-only articles do not make a page positive. Use acquisition evidence and the scope’s wording to describe the supporting-information situation; local absence does not prove nonexistence.

4. Audit the paper page

   Compare the bibliography metadata and every retained caption, identifier, label, value, qualifier, and locator with the source in a second pass. Check that no eligible record or necessary qualifier was skipped and that exclusions follow the reported result rather than the surrounding experiment’s purpose. Have a source-aware reviewer who did not transcribe the page challenge coverage, scope decisions, transcription, record grouping, and provenance; fix every substantiated defect.

5. Update the dated index

   Follow the established dated-index format. The dated index must contain exactly one row for every sibling paper page and no other rows. Derive each link, displayed title, and analysis-availability cell only from the retained paper page; count only transcribed analytical results, not coverage notes, identifiers, or acquisition settings alone. Do not infer availability from the source documents. When creating a dated index, use an existing dated index as the structural reference and add its YYYY-MM-DD/ directory link to ext/index.html in ascending date order.

6. Validate the change

   During paper processing, run:

   python3 scripts/validate_characterization_html.py ext/YYYY-MM-DD/MAIN-SOURCE-STEM.html

   Before completing a date, run:

   python3 scripts/validate_characterization_html.py ext/YYYY-MM-DD/

   Fix the underlying defect behind every reported problem; do not bypass or game the validator by suppressing diagnostics, weakening required content or structure, or changing source data merely to pass. Review the dated index manually for one-to-one paper coverage and accurate titles and analysis cells. Verify the date link in ext/index.html. Run git status and inspect the diff to ensure it contains only the intended paper pages, dated index, and parent-index link, with no extraction artefacts. Do not commit or publish unless requested.

Completion

A date is complete when every acquired source has been inspected completely, every acquired main paper has an audited page, every paper page has exactly one accurate dated-index row, the dated index is linked from ext/index.html, and the final validation command passes.