The metric data sharing has been waiting for

DOIphin, our federated aggregator, cross-walks each DOI across 14+ scholarly APIs (CrossRef, OpenAlex, DataCite, Zenodo, Dryad, and more) into one unified record — and builds the citation/link graph DataRank scores. FAIR/DataCite signals are kept as context.

Our DrPaper classifier — a fine-tuned SciBERT model — reads each paper and decides whether its main contribution is a dataset (cohort, atlas, benchmark, database) versus a method, theory, or review. Only data papers receive a percentile ranking.

Each paper is evaluated against the FAIR principles (Findable, Accessible, Interoperable, Reusable). Shown alongside scores for transparency — not part of the score itself.

For each paper we fetch its citers from OpenAlex — the papers that cite it — to measure downstream influence.

Combine the paper's own citation count with a one-step propagation through its citers, weighted so heavily-cited citers count more. Self-citations are removed.

Sort data papers by DataRank and map each to a percentile in [0, 100] within the data-paper corpus. A paper at the 99th percentile is in the top 1% by impact.