Datasetpublic

cometadata/crossref-arxiv-citations

Crossref arXiv Citations A dataset of arXiv preprints and their citations extracted from Crossref metadata, validated against DataCite records. Dataset Description This dataset maps arXiv works to the works in Crossref that cite them. Each record represents an arXiv preprint with all known citations from Crossref-registered works. Built from the Crossref Metadata Plus monthly snapshot 2026-07 and the DataCite monthly data file 2026-07 with… See the full description on the dataset page: https://huggingface.co/datasets/cometadata/crossref-arxiv-citations.

sourceHugging Facecc0-1.0updated 4d agoView on Hugging Face

0likes102downloads

Dataset Card

Crossref arXiv Citations

A dataset of arXiv preprints and their citations extracted from Crossref metadata, validated against DataCite records.

Dataset Description

This dataset maps arXiv works to the works in Crossref that cite them. Each record represents an arXiv preprint with all known citations from Crossref-registered works.

Built from the Crossref Metadata Plus monthly snapshot 2026-07 and the DataCite monthly data file 2026-07 with crossref-citation-extraction version 2.2.0.

Dataset Configurations

Config	Description	arXiv Works	Citing Works	Reference Matches
`all`	All validated citations (default)	1,042,011	5,999,691	6,028,756
`asserted`	Citations where the DOI was explicitly provided by the publisher or matched by Crossref	160,091	452,505	454,190
`mined`	Citations mined from reference metadata without a publisher or Crossref DOI assertion	998,112	5,548,053	5,574,566

Loading the Dataset

python

from datasets import load_dataset

dataset = load_dataset("cometadata/crossref-arxiv-citations")
asserted = load_dataset("cometadata/crossref-arxiv-citations", "asserted")
mined = load_dataset("cometadata/crossref-arxiv-citations", "mined")

Data Schema

Each record contains:

Field	Type	Description
`arxiv_doi`	string	The DOI for the arXiv work (format: `10.48550/arXiv.{id}`)
`arxiv_id`	string	The arXiv identifier (e.g., `1412.6980`)
`reference_count`	integer	Total number of reference match instances
`citation_count`	integer	Number of unique citing works
`cited_by`	array	List of citing work objects

Each citing work object in cited_by contains:

Field	Type	Description
`doi`	string	DOI of the citing work
`provenance`	string	Highest-confidence provenance among this work's matches (`publisher` > `crossref` > `mined`)
`matches`	array	Reference matches from this citing work

Each match contains:

Field	Type	Description
`provenance`	string	How this match was obtained (see below)
`raw_match`	string	The exact reference text that identified the arXiv work (a substring, or a whole structured value): an arXiv identifier, a DOI, or a URL such as the work's registered landing page
`reference`	struct	The Crossref reference metadata (string-valued fields; absent keys are null)

Provenance Values

—publisher - the reference's structured DOI field holds the arXiv DOI with doi-asserted-by: publisher
—crossref - the reference's structured DOI field holds the arXiv DOI with doi-asserted-by: crossref
—mined - any other match: an arXiv identifier, DOI, or URL found in the reference's DOI, URL, article-title, journal-title, or unstructured field (a structured DOI counts as mined unless doi-asserted-by is publisher or crossref)

The asserted config contains only publisher and crossref matches; the mined config contains only mined matches. A citing work appears in a config only with its matches of that config's provenances, and all counts are recomputed per config.

Extraction Process

arXiv references are extracted from the Crossref snapshot by scanning reference metadata for arXiv identifiers in modern (arXiv:2403.03542), legacy (arXiv:cs.DM/9910013), DOI (10.48550/arXiv.2403.03542), and URL (arxiv.org/abs/2403.03542) forms. Identifiers are normalized (lowercase, version suffixes removed). Extracted arXiv DOIs are validated against an index of DOIs built from the DataCite monthly data file 2026-07; only arXiv works registered in DataCite are included. When the dataset is built by the full DataCite pipeline, an arXiv work can also be matched by its DOI anywhere in the reference or by its registered DataCite landing-page URL.

Data Sources

—Crossref Metadata Plus monthly snapshot 2026-07
—DataCite monthly data file 2026-07
—Extraction tools: cometadata/crossref-arxiv-citation-extraction

Limitations

—Only includes citations registered in Crossref metadata
—Reference parsing depends on successful identifier extraction from reference metadata
—The mined subset may contain lower-quality matches from noisy reference text

License

This dataset is released under CC0 1.0 Universal.

Citation

bibtex

@dataset{crossref_arxiv_citations,
  title = {Crossref arXiv Citations},
  author = {Cometadata},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/cometadata/crossref-arxiv-citations}
}