The XML query
Which DOI held here is this citation? Ask with the journal, ISSN, volume, issue, first page, year, author or title, in Crossref's XML query form; the answer comes in Crossref's XML result form.
What it does
A reference list, a reference manager or an index holds citations without DOIs. Send each as a query and the registry answers with the DOI it holds for it — resolved (one match), multiresolved (several; the query is ambiguous unless you ask for every hit), unresolved (none) or malformed (nothing to match on). Each field in a resolved answer says how it matched: exact, fuzzy, or optional (given, but not needed to match). Only the records held in this registry are searched; a DOI registered elsewhere is not found here.
The forms are Crossref's: a query_batch (the query schema, examples) in, a crossref_result (crossref_query_output3.0.xsd) out. Fields: doi, issn, isbn, journal_title, volume_title, series_title, proceedings_title, author (with search-all-authors), ORCID, volume, issue, first_page, year, article_title, unstructured_citation, identifier; each may carry match="fuzzy|exact|optional|null". Query attributes: key, enable-multiple-hits, expanded-results (every author, not only the first), secondary-query="author-title" (tried when the full query finds nothing). No e-mail address or account is needed; one given as pid= is echoed back.
Try a query
Or the fields, one query
For programs
curl -H "Content-Type: application/xml" --data-binary @query.xml "https://registry.smartscholars.in/query.php"
curl "https://registry.smartscholars.in/query.php?issn=2456-5660&volume=10&issue=12&first_page=35&year=2025"
Every answer is text/xml with HTTP 200 — an unresolved query is an answer, not a failure; a request that cannot be read (not XML, no query_batch, no query) is 400 with the reason in XML. format=unixref or format=unixsd gives the matched records whole instead of the fields. A DOI-to-metadata query — a query holding only <doi> — works the same way. The same record, one at a time, is also the resolver's by content negotiation, and the documentation says how the match is made. Reads are counted like every other door's (keys and limits).



