Graph view

The citation graph, as a map

Citation indexes flatten the literature into a number. This is the shape of what CiteStamp has recorded so far — sweeps of SSRN and Zenodo metadata, plus every signed claim. Enter an identifier — a DOI, a Zenodo record, a preprint — and walk the neighborhood node by node. Signed claims draw in solid gold; machine-inferred edges draw in dashed steel. The two never mix — here or anywhere else on CiteStamp. A sparse neighborhood means the sweeps have not reached a work’s citers yet — not that nobody cites it.

Start here. Every dot is a real work and every line a recorded claim. Click a dot and a few of its own connections come with it.

signed inferred real edges around 10.5281/zenodo.18262424

A few dots at a time, on purpose. To walk the whole neighborhood, with every label and every hop, open it in the full map. Or map a work of your own.

The map takes an identifier, not a title: a DOI, an arXiv id or a Zenodo record. A paper’s DOI is on its publisher or repository page, and /check resolves a pasted reference to its DOI.

Want all of it? The inferred graph CiteStamp builds — 15M edges, CC0 — is deposited at Harvard Dataverse; the signed log is mirrored on GitHub. The openalex-snapshot rows are projected from OpenAlex, which publishes them at source.

What you are looking at. Every line is a recorded claim with an attribution: a named researcher who signed it, or the open-metadata sweep that inferred it. Nothing here is a verdict — it is a record of who said what, and how it got on the record. Node labels are titles fetched from Crossref and DataCite so the map can be read; a title is public metadata about a work, never part of the signed record, and a work whose registry has no title for it is drawn as its identifier.