Structured full text
JATS XML conversion, without a conversion vendor
JATS is the standard form of a journal article: the text with its parts labelled, so that a machine knows which words are the abstract, which are a figure caption, which author belongs to which affiliation, and what each reference points at. It is what abstracting and indexing services, archives and funders expect to be given.
It is also the file most journals cannot make themselves, so it is bought by the article from a conversion house - which is why it usually arrives after publication rather than before, and why nobody looks at it closely.
The difficulty
Why it is harder than it sounds
- A manuscript says how it looks, not what it is
- An author makes a heading by setting a line in bold. A converter sees bold text; the standard needs a heading, at a level, containing a section. Everything downstream - the table of contents, the running heads, the navigation online - depends on that distinction being got right.
- Valid is not the same as correct
- A file can satisfy the schema and still be wrong: an affiliation attached to the wrong author, a figure in the wrong section, a reference whose year does not match the work it names. Validation catches the first kind of error and is silent about the second.
- The hard parts are the structured ones
- Tables with merged cells, multi-level lists, equations, footnotes that sit inside a table cell, and references in a style nobody has seen before. These are where hand conversion gets expensive and where cheap conversion gets quietly lossy.
- It has to agree with the PDF
- The XML and the galley describe one article. If they are made by two different processes at two different times, they drift - and the version of record disagrees with the version your index has.
The approach
How it works here
The XML is not a separate deliverable produced afterwards. It is the same article the editor approved, written out in another form.
-
The structure is proposed, not assumed
The manuscript is read and its parts identified - sections and their levels, authors and affiliations, figures and their captions, tables, equations, footnotes and references.
-
An editor confirms it
That structure is shown as a document you can correct. This is the step a conversion vendor performs out of your sight and cannot ask you about; here the person who knows the article does it.
-
It is checked before it is accepted
The article is validated against the standard, and against the things the standard cannot express - a reference with no year, an author with no affiliation, a figure nobody cites. You are told what is missing rather than handed a file that is merely well-formed.
-
Every output is written from the same approved article
The galley, the online version, the XML and the deposit metadata all come from one source, at one moment, so they cannot disagree with each other.
What survives
What is carried through
- Sections and subsections at their proper levels, including appendices
- Authors, their affiliations, their ORCIDs and the corresponding author
- Abstract, keywords, and the article’s own licence and copyright line
- Figures with captions, numbering and the citations that refer to them
- Tables with header rows, merged cells and footnotes
- Equations as mathematics, not as pictures of mathematics
- Footnotes, cross-references and competing-interest statements
- A reference list, parsed and matched, in the journal’s citation style
And the deposit that registers your DOI, with the metadata in the form the registry expects - so the registration can go out the day you publish rather than whenever the conversion comes back.
Going deeper
Read more on this
- What is JATS XML? What JATS XML is, what goes inside one, why a valid file is not the same as a correct one, and which services ask journals for it. Read this
- Word to JATS XML Why a Word file is hard to convert, where conversions quietly lose tables and equations, and why a person has to be in the middle of it. Read this
- PubMed Central deposit What a PubMed Central deposit consists of, the metadata it demands, where deposits get rejected, and why to produce it at publication. Read this
- Crossref DOI registration What a DOI registration contains, how to choose the rest of the name, the obligation it creates, and the metadata mistakes that recur. Read this
- XML-first publishing Why PDF-first lets the deposit and the published article diverge, what XML-first changes in practice, and how to move one journal to it. Read this
Read next
-
Journal typesetting software
What journal typesetting software has to do, what to look for when choosing it, and how an editor produces a print-ready galley without a typesetter.
Read this -
File formats
Every file produced from one manuscript: a typeset PDF, online HTML, an editable Word copy, EPUB, JATS XML and the Crossref, DOAJ and PubMed deposits.
Read this -
For society journals
Run a society journal without a production department: an editor uploads the accepted paper and the issue is typeset, deposited and online the same day.
Read this
Next step
See what it makes of one of your articles.
Send us a paper you have already published, with your journal's details. We set the account up and you compare the result with what you paid for, which is the fastest way to judge it.
