Preserve Identifier Namespaces in Article XML

Extract every typed identifier

An article record may include publisher identifiers, a DOI, and a database-specific identifier. Inspect the article front matter and save each article-id with the identifier type and source document. Scope the extraction to the article being described. A DOI in a reference list identifies a cited object and can contaminate the article key if a parser simply takes the first matching string anywhere in the file.

Keep type, raw text, and normalized value in separate fields. Apply normalization according to the identifier scheme. If the type is absent or unfamiliar, preserve the value for review. A string that happens to contain digits does not reveal which organization assigned it.

Check a hypothetical collision

Two articles each carry three identifiers, giving six identifier rows. Each publisher uses a local number 104 for its own article, while the remaining four identifiers differ. Grouping by raw value yields five values and an apparent cross-article match. Including the issuing publisher in the local identifier's namespace recovers the two distinct article identities.

Test the import on documents containing cited DOIs, supplementary identifiers, and nested article sections. Verify the association between each identifier and its owning article. Downstream joins should retain the original article-row count unless the query explicitly requests one row per identifier.

JATS article identifier reference

Related worksheet: Table grain and primary-key worksheet.

Source checked on 5 November 2026.