Neatbo.

Why matching citation counts can hide a failed migration

Check every author, title character, identifier and sidecar-only field; an unchanged reference count can conceal lost citation meaning.

Counts do not prove a complete citation

A converted file can have the same number of references while its second author, DOI, Greek φ, or title case has changed. Each is a meaningful defect for a literature handoff. The public discussion behind this workflow includes a researcher using ris2xml followed by xml2bib, and another reporting that a web converter accepts concatenated RIS entries. The original question supplied JSON and eventually found an existing BibTeX entry; it did not supply the test RIS bytes used here.

A matching record count can miss these differences
Item to compareWhat counts can hideWhere to check
Ordered authorsA missing second author or a changed suffix can leave the record count unchanged.Compare all names, their order and corporate spelling in the complete target and field report.
Title and φCharacters or case protection can change while the entry remains readable.Compare interpreted target text with the original braces and commands retained in citation-mapping.json.
DOI and serial identifiersA changed DOI, ISBN or ISSN still belongs to one record.Compare the full values and the record type; RIS SN maps to issn for JOUR and isbn otherwise.
Unfamiliar metadataA target file can omit an unrepresentable field without losing an entry.Use default rejection or explicitly keep the complete sidecar; sidecar-only fields have no target representation.

Keep representation differences visible

BibTeX names with suffixes are ordered family, suffix, given. RIS specifies family, given, suffix. Corporate-author comma conventions and brace protection also differ. Moving strings unchanged can produce a syntactically readable file with different meaning.

A field report makes those transformations inspectable. Unknown metadata belongs in an explicit complete source report when it cannot be represented in the target. A successful conversion is an opportunity to check that report, rather than a reason to discard it.

  • Select BibTeX → RIS or RIS → BibTeX explicitly. This workflow does not convert NBIB or JSON and does not fetch DOI metadata.
  • Macro definitions/references, preambles and cross-references are refused even with sidecar mode. Prepare a resolved export in the original manager and keep the original.
  • Save both the complete target citation file and citation-mapping.json. Keep every sidecar-only field with its original text and source position.
  • Import a copy into the intended manager and inspect all fields. Corporate trailing-comma conventions and title case handling can differ between managers.

Read the whole target independently

The staged native evidence reads article, book and conference-paper outputs with independent bibliography readers and compares every field. It also alters physical output files by removing an author, changing φ and DOI, or damaging braces, and confirms that independent validation detects each change. Those checks demonstrate the local format work; browser interaction, receiving software behavior and formal admission still need their own review.

References

  • Citation migration question and replies

    The question supplied JSON and later found an existing BibTeX entry. Replies describe an actual ris2xml/xml2bib workflow and a web converter accepting concatenated RIS; they do not supply the test RIS bytes used here.

  • Original BibTeX manual

    Explains original BibTeX entries, macros, cross-references and title case protection. This workflow refuses macros and cross-references and retains exact source braces in its report.

  • Citation.js parser documentation

    Documents Citation.js raw BibTeX parsing, field value types and LaTeX interpretation. It supports the parser choices here and does not establish a receiving manager’s import behavior.