Sci-database LogoSci-database

A DOI-tracked export: downloadable worked example

Documentation example v1.0 · Prepared 2 September 2026 · Sci-database

This small example shows how to keep an exported row connected to its source. It is a curated bibliographic fixture, not a live AI extraction. No AI Judge run or human research audit is recorded, and no accuracy percentage is claimed.

One source, one inspectable row

Title
The FAIR Guiding Principles for scientific data management and stewardship
Publication
Scientific Data, 2016, volume 3, article 160018
Source-paper DOI
10.1038/sdata.2016.18
Evidence location
Publisher article header and “Cite this article” block. No PDF page number or coordinates are asserted.

Download the example files

Public downloads work without signing in. CSV and JSON are generated from the same fixture with the current plotting-export builder. Keep the files together.

Reproduce the checks

  1. Open the CSV in your plotting tool, or parse the JSON. Confirm there is one row and retain record_id and source_doi.
  2. Compare the four bibliographic fields with the publisher citation block. Consult the data dictionary for the added tracking columns.
  3. Read provenance.json: it explicitly records that no AI extraction or Judge run occurred. The builder’s generic ai_extracted status means no validation evidence is recorded; it is not a claim about how this fixture was prepared.
  4. Read checks.json: one record, zero validation coverage and an unknown agreement score. The example does not satisfy marketplace publication minimums.
  5. Compare downloaded file SHA-256 hashes with manifest.json. On Windows use Get-FileHash data.csv -Algorithm SHA256. Hashes detect byte changes, not truth.

What this example does not establish

There are no experimental measurements, error bars, extracted table values or performance results here. A successful import is not evidence of scientific accuracy. No dataset license or dataset DOI is assigned. Source-paper rights are separate from rights to a derived release.

Read the methodology before using extracted data in research.