Metadata Profile 1.0
How a record here holds a title, an abstract and a name in every script it was written in — and what is refused, what is only warned, and what each export carries. Version 1.0, 19 October 2026, for records in format 1.2 (smartscholars-doi-metadata/1.2).
The principle. A title in its own script, the same title in Latin letters, and its translation are equal titles: each is kept whole, each says its language and script in a BCP 47 tag, and each says what it is — so a reader who types the Latin form, an index that wants English, and a library that catalogues in the original all find the same record. Every list and every answer on this page is read from the registry's own code when the page opens; the round-trip report shows the same rules followed through every export (last run 2026-10-19: every step passed).
Titles
Each title is an object: value, type, lang, and — new in 1.2 — role. The record's own title is the PrincipalTitle; every other title stands beside it.
| Field | Holds |
|---|---|
type | AbbreviatedTitle, DistinctiveTitle, FormerTitle, Name, PrincipalTitle, Title, TranslatedTitle — the DOI Kernel's own names (Attribute Value Sets 2.3) |
lang | a BCP 47 language tag — te, hi, ta; with the script when the text is not in the language's own script: te-Latn, hi-Latn, ur-Latn |
role | transliteration or translation: transliteration — the same words in another script (its tag says the script); translation — the title in another language |
original | true on the title in the original language of a translated work (format 1.1) — never together with a role |
An example, the Telugu sample: “తెలంగాణ గ్రామాల్లో భూగర్భ జలాల నాణ్యత” te principal · “Telangana gramallo bhugarbha jalala nanyata” te-Latn transliteration · “Groundwater quality in the villages of Telangana” en translation ·
Names
A person is family and given (an organisation org; a name that cannot be split, display). New in 1.2, name.alt lists the same name in other languages or scripts — at most 5 forms, one a language or script, each with its lang — so an author is found under the name on the printed page and the name on the record.
| Role in the record | The DOI Kernel's agent role |
|---|---|
author | Author |
editor | Creator |
translator | Creator |
contributor | Creator |
chair | Creator |
corporate_author | CorporateCreator |
publisher | Publisher |
funder | FundingBody |
Languages and scripts — India's scheduled languages
A text of three letters or more, tagged with a language, must hold at least one letter of that language's script — or the record is refused, with the fix. The script each tag is checked against, as the registry reads it:
| Language | Tag | Checked against | In Latin letters |
|---|---|---|---|
| Assamese | as | Bengali | as-Latn |
| Bengali | bn | Bengali | bn-Latn |
| Bodo | brx | written in more than one script — checked only when the tag names it: brx-Deva | brx-Latn |
| Dogri | doi | Devanagari | doi-Latn |
| Gujarati | gu | Gujarati | gu-Latn |
| Hindi | hi | Devanagari | hi-Latn |
| Kannada | kn | Kannada | kn-Latn |
| Kashmiri | ks | written in more than one script — checked only when the tag names it: ks-Arab or ks-Deva | ks-Latn |
| Konkani | kok | Devanagari | kok-Latn |
| Maithili | mai | Devanagari | mai-Latn |
| Malayalam | ml | Malayalam | ml-Latn |
| Manipuri | mni | written in more than one script — checked only when the tag names it: mni-Beng or mni-Mtei | mni-Latn |
| Marathi | mr | Devanagari | mr-Latn |
| Nepali | ne | Devanagari | ne-Latn |
| Odia | or | Oriya | or-Latn |
| Punjabi | pa | Gurmukhi | pa-Latn |
| Sanskrit | sa | Devanagari | sa-Latn |
| Santali | sat | Ol_Chiki | sat-Latn |
| Sindhi | sd | written in more than one script — checked only when the tag names it: sd-Arab or sd-Deva | sd-Latn |
| Tamil | ta | Tamil | ta-Latn |
| Telugu | te | Telugu | te-Latn |
| Urdu | ur | Arabic | ur-Latn |
What is refused, and what is only warned
Each line is a small change to the Telugu sample, and the answer is the registry's own, worked out when this page opened. Refused: the record is not registered until it is put right. Warned: registered as sent, and the warning comes back with it.
| The rule | The change tried | The registry's words | |
|---|---|---|---|
| A title's role is "transliteration" or "translation" | titles[1].role: "alias" | Refused | titles[1]: role must be one of transliteration, translation |
| The principal title has no role: it is the title of the item registered | titles[0].role: "translation" | Refused | titles[0]: the PrincipalTitle is the title of the item registered — a translation is another title beside it |
| A title is the original-language title of a translated work or a transliteration or translation — not both | titles[2].original: true | Refused | titles[2]: a title is the original-language title or a translation, not both |
| A title with a role has a language tag | titles[1] without lang | Refused | titles[1]: a transliteration needs its lang — a BCP 47 tag with the script it is written in (hi-Latn, ur-Latn, ta-Latn) |
| A transliteration's tag names its script | titles[1].lang: "te" (for te-Latn) | Refused | titles[1]: a transliteration says its script in the tag — te-Latn (or -Deva, -Arab, -Taml …), not "te" alone |
| A translation is in another language than the principal title | titles[2].lang: "te" | Refused | titles[2]: a translation is in another language than the principal title (te) — for the same words in another script use role transliteration |
| A name's other forms are a list of forms | agents[0].name.alt: "రెడ్డి" | Refused | agents[0].name.alt lists the name in other languages or scripts: [{display | family (and given) | org, lang}] |
| Each other form of a name has its language tag | agents[0].name.alt[0] without lang | Refused | agents[0].name.alt[0]: lang is required — a BCP 47 tag, with the script when it is a transliteration (hi-Latn) |
| One form of a name a language or script | two forms tagged "te" | Refused | agents[0].name.alt[1]: a second form in "te" — one form a language or script |
| At most 5 other forms of a name | six forms | Refused | agents[0].name.alt: at most 5 other forms of a name |
| No replacement character — a letter lost when the text was converted | U+FFFD added to titles[0] | Refused | titles[0]: holds the replacement character � (U+FFFD) — a letter was lost when this text was converted (a legacy font, OCR); retype the text from the printed article |
| No private-use character — a legacy font's own code | U+E000 added to titles[0] | Refused | titles[0]: holds a private-use character (U+E000) — a legacy font's own code that no one else can read; retype the text in Unicode |
| No control character — the sign of text read in the wrong encoding | U+0096 inside titles[2] | Refused | titles[2]: holds a control character (U+0096) — the sign of Windows-1252 text read as Latin-1; send the text as UTF-8 from its source |
| The tag's script appears in the text | titles[0]: Latin text tagged "te" | Refused | titles[0]: is tagged te (Telugu) but has no Telugu letter — it is written in Latin; tag the language it is written in, or te-Latn if it is te written in Latin letters (a transliteration) |
| No word written in two alphabets — a look-alike letter makes it unfindable | a Cyrillic "а" inside "Groundwater" | Refused | titles[2]: the word "Groundwаter" mixes Latin and Cyrillic letters — a look-alike letter from another alphabet (often from a PDF copy) makes the word unfindable; retype it |
| Greek letters inside a Latin word stay allowed (μm, αKG) | titles[2]: "Particles of 5 μm in groundwater" | Accepted | accepted — nothing to say |
| Text in Unicode NFC (composed) form | a decomposed "é" in titles[2] | Warned | titles[2]: not in Unicode NFC (composed) form — search and matching read composed letters; send it normalised (NFC) |
| No tab or line break inside a title or a name | a line break inside titles[2] | Warned | titles[2]: a tab or line break inside a title — write it on one line |
| No invisible byte-order mark | U+FEFF before titles[2] | Warned | titles[2]: holds an invisible byte-order mark (U+FEFF) — take it out |
What each export carries
Every format a record is given out in, and what it holds of the three titles, a second-language abstract and the author's name in its own script. "Not written" is a format with no place for it, or this registry writing the principal title alone into that form. The round trip checks every export against exactly this table.
| Export | the title in its own script | its transliteration | its translation | the abstract in a second language | the author's name in its own script |
|---|---|---|---|---|---|
| the record (JSON) | with its tag | with its tag | with its tag | with its tag | with its tag |
| DOI Kernel Metadata Declaration (XML) | with its tag | with its tag | with its tag | not written | not written |
| DataCite XML (Metadata Schema 4.7) | with its tag | with its tag | with its tag | with its tag | not written |
| RDF (Turtle) | with its tag | with its tag | with its tag | with its tag | not written |
| RDF/XML | with its tag | with its tag | with its tag | with its tag | not written |
| schema.org in JSON-LD | with its tag | without a tag | without a tag | not written | not written |
| Crossref UNIXREF XML | with its tag | not written | not written | with its tag | not written |
| Crossref UNIXSD XML | with its tag | not written | not written | with its tag | not written |
| CSL-JSON | with its tag | not written | not written | not written | not written |
| BibTeX | with its tag | not written | not written | not written | not written |
| RIS | with its tag | not written | not written | not written | not written |
| a formatted reference, APA (7th) | without a tag | not written | not written | not written | not written |
DataCite XML says each title's role in its own words — a transliteration is an AlternativeTitle, a translation a TranslatedTitle. The record itself (JSON, and the API) always holds everything.
Record types
| Type | What it is |
|---|---|
JournalArticle | an article in a journal |
Journal | a journal as a whole (title-level DOI) |
JournalIssue | one issue of a journal |
JournalVolume | one volume of a journal |
Book | a book (monograph, edited volume) |
BookChapter | a chapter or part of a book |
BookSeries | a series of books |
ConferenceProceedings | a conference paper or a proceedings volume |
Dataset | a dataset or a database record |
Report | a report |
Thesis | a thesis or dissertation |
Standard | a standard |
Figure | a figure, table or supplementary file |
WebResource | a web resource or a posted content item (preprint, blog post) |
The format: format 1.2 in the documentation · its JSON Schema · examples in six languages · the round-trip report. Standards it follows: BCP 47 language tags (RFC 5646), ISO 15924 script codes, Unicode Normalization Form C.
