{"success":true,"data":{"author":"COMET (cometadata)","bucketAccessType":"public","bucketName":"cometadata/crossref-arxiv-citations","bucketPrefix":"","category":"Open Data","cloudProvider":"huggingface","createdAt":"2026-11-21T22:16:38.631Z","description":"A citation graph linking arXiv preprints to the Crossref-registered works (journal articles, proceedings, books) that cite them. The citations were extracted from the reference lists in the Crossref metadata snapshot of 2026-07 and validated against the DataCite data file of the same month, using the crossref-citation-extraction tool v2.2.0. The default configuration covers 1,042,011 cited arXiv works, 5,999,691 citing works and 6,028,756 reference matches. Each record gives the arXiv DOI and id, the reference and citation counts, and the list of citing DOIs, each with its provenance. The provenance values are publisher-asserted, Crossref-matched, or mined from reference text such as an arXiv id, DOI or URL, and the raw matched string and the Crossref reference metadata come with each match. The asserted and mined configurations separate high-confidence links from mined ones. Bibliometrics and science-of-science researchers use it to measure how preprints are cited in the published literature, which standard indexes undercount. The data is Parquet under data/; load it with Hugging Face datasets or pyarrow. Released by COMET under CC0 1.0.","displayTitle":"Crossref arXiv Citations: 6 Million Journal-Article Citations to 1 Million arXiv Preprints (COMET)","huggingFaceUrl":"https://huggingface.co/datasets/cometadata/crossref-arxiv-citations","id":"2bd228a2-9d50-53fa-b84a-9e28818e638c","license":"CC0-1.0","licenseEvidenceUrl":"https://huggingface.co/datasets/cometadata/crossref-arxiv-citations","objectCountEstimate":6,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"cometadata/crossref-arxiv-citations\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"cometadata/crossref-arxiv-citations\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=cometadata, license:cc0-1.0, size_categories:1M<n<10M, format:parquet, downloads=216"],"format":"mixed","generatedAt":"2026-11-21T22:31:55.409Z","layout":{"example":"data/all.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/all.parquet","truncatedAt":512}],"sampleObjects":["data/all.parquet","data/asserted.parquet","data/mined.parquet","metadata/pipeline-manifest.json"],"sidecars":[{"bytes":"---\nlicense: cc0-1.0\ntask_categories:\n  - text-classification\n  - feature-extraction\nlanguage:\n  - en\ntags:\n  - arxiv\n  - crossref\n  - citations\n  - preprints\n  - scholarly-communications\n  - bibliometrics\n  - doi\npretty_name: Crossref arXiv Citations\nsize_categories:\n  - 1M<n<10M\nconfigs:\n  - config_name: all\n    data_files:\n      - split: train\n        path: data/all.parquet\n    default: true\n  - config_name: asserted\n    data_files:\n      - split: train\n        path: data/asserted.parquet\n  - config_name: mined\n    data_files:\n      - split: train\n        path: data/mined.parquet\n---\n\n# Crossref arXiv Citations\n\nA dataset of arXiv preprints and their citations extracted from Crossref\nmetadata, validated against DataCite records.\n\n## Dataset Description\n\nThis dataset maps arXiv works to the works in Crossref that cite them. Each\nrecord represents an arXiv preprint with all known citations from\nCrossref-registered works.\n\nBuilt from the Crossref Metadata Plus monthly snapshot `2026-07` and the\nDataCite monthly data file `2026-07` with\n[crossref-citation-extraction](https://github.com/cometadata/crossref-arxiv-citation-extraction)\nversion 2.2.0.\n\n### Dataset Configurations\n\n| Config | Description | arXiv Works | Citing Works | Reference Matches |\n|--------|-------------|-------------|--------------|-------------------|\n| `all` | All validated citations (default) | 1,042,011 | 5,999,691 | 6,028,756 |\n| `asserted` | Citations where the DOI was explicitly provided by the publisher or matched by Crossref | 160,091 | 452,505 | 454,190 |\n| `mined` | Citations mined from reference metadata without a publisher or Crossref DOI assertion | 998,112 | 5,548,053 | 5,574,566 |\n\n### Loading the Dataset\n\n```python\nfrom datasets import load_dataset\n\ndataset = load_dataset(\"cometadata/crossref-arxiv-citations\")\nasserted = load_dataset(\"cometadata/crossref-arxiv-citations\", \"asserted\")\nmined = load_dataset(\"cometadata/crossref-arxiv-citations\", \"mined\")\n```\n\n## Data Schema\n\nEach record contains:\n\n| Field | Type | Description |\n|-------|------|-------------|\n| `arxiv_doi` | string | The DOI for the arXiv work (format: `10.48550/arXiv.{id}`) |\n| `arxiv_id` | string | The arXiv identifier (e.g., `1412.6980`) |\n| `reference_count` | integer | Total number of reference match instances |\n| `citation_count` | integer | Number of unique citing works |\n| `cited_by` | array | List of citing work objects |\n\nEach citing work object in `cited_by` contains:\n\n| Field | Type | Description |\n|-------|------|-------------|\n| `doi` | string | DOI of the citing work |\n| `provenance` | string | Highest-confidence provenance among this work's matches (`publisher` > `crossref` > `mined`) |\n| `matches` | array | Reference matches from this citing work |\n\nEach match contains:\n\n| Field | Type | Description |\n|-------|------|-------------|\n| `provenance` | string | How this match was obtained (see below) |\n| `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 |\n| `reference` | struct | The Crossref reference metadata (string-valued fields; absent keys are null) |\n\n### Provenance Values\n\n- `publisher` - the reference's structured `DOI` field holds the arXiv DOI with `doi-asserted-by: publisher`\n- `crossref` - the reference's structured `DOI` field holds the arXiv DOI with `doi-asserted-by: crossref`\n- `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`)\n\nThe `asserted` config contains only `publisher` and `crossref` matches; the\n`mined` config contains only `mined` matches. A citing work appears in a\nconfig only with its matches of that config's provenances, and all counts are\nrecomputed per config.\n\n## Extraction Process\n\narXiv references are extracted from the Crossref snapshot by scanning\nreference metadata for arXiv identif","encoding":"utf8","key":"README.md","name":"README.md","size":5497,"truncated":true}],"sourceObjectKey":"data/all.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"cometadata/crossref-arxiv-citations\")\nprint(ds)","starterSamplePath":"data/all.parquet","totalObjects":6},"summary":"A citation graph linking arXiv preprints to the Crossref-registered works (journal articles, proceedings, books) that cite them.","title":"Crossref arXiv Citations: 6 Million Journal-Article Citations to 1 Million arXiv Preprints (COMET)","updatedAt":"2026-11-22T01:08:32.316Z","visibility":"public","websiteUrl":"https://github.com/cometadata/crossref-arxiv-citation-extraction"}}