map the research economy.
research-atlas reconciles the world’s public research funding into one normalized graph — funders, grants, organizations, people, works and fields — keyed on global identifiers (ROR for orgs, ORCID for people, OpenAlex/DOI for works), with full provenance on every row. It is the graph the funding-landscape paper is built on, and the source of every open dataset on Bucket.
six entities, one graph
Funder → Grant → Organization → Person → Work → Field, joined by directed edges (grant_work, person_org, work_field, …). The graph passes 36/36 referential-integrity checks with zero orphan edges.
NIH (+ awarding IC), NSF, EC/ERC, UKRI, Gates, Wellcome, Sloan, DFG — 73 in all.
Every award record, USD-normalized with a stamped FX date and full provenance.
Recipient institutions, merged to one canonical node per ROR id.
Investigators, keyed on ORCID where available (61% coverage).
Research outputs that acknowledge a funder, linked via OpenAlex.
The OpenAlex topic taxonomy, joined to works by work_field edges.
query the atlas
A live, paid-per-query API over the full ~8.1M-row graph is on the roadmap — ask “who funds CRISPR delivery?” or “which institutions co-publish with the ERC?” and get a cited answer over feed402/x402. It is not wired yet.
Until the live API lands, the full graph is available as open datasets (a 892-row sample manifest is published now) and as the reproducible pipeline on GitHub.