rxiv

about api
find the newest relevant papers from all preprint servers.

what is rxiv?

rxiv scans 242 preprint sources — including bioRxiv, medRxiv, arXiv, and 20+ others — and finds papers relevant to your research. just search your name.

how it works

we look up your publications on Semantic Scholar and build an embedding profile using OpenAI's text-embedding-3-small. every preprint in our database is also embedded. we then rank preprints by how similar they are to your publication profile using cosine similarity.

the ranking combines three signals:

centroid — the average of your publication embeddings. represents your current research focus.
velocity — the direction your publications are moving in embedding space over time. surfaces papers aligned with where your research is heading, not just where it's been.
gap — midpoints between clusters of your publications. finds papers in the intersection of your interests that you might not search for directly.

these are weighted 50/30/20 and combined into a single score. the top 100 papers are returned.

sources

we pull from 242 sources across 231 RSS feeds and 11 API providers, updated twice daily:

bioRxiv (26 categories) · medRxiv (51) · arXiv (144 — physics, CS, math, q-bio, econ, stats) · PsyArXiv · SocArXiv · EdArXiv · EarthArXiv · EcoEvoRxiv · engrXiv · NutriXiv · PaleorXiv · MetaArXiv · SportRxiv · Research Square · F1000Research · NBER · MPRA · and more.

multiple researchers

you can add more than one researcher. their publications are pooled together before computing the profile, so the results reflect the combined interests of a lab or research group.

bookmarkable URLs

your researchers are saved in the URL: ?a=3128227,2157025857. bookmark it, share it, or put it on a lab screen. results update every time you visit.

similar researchers

after results load, rxiv shows researchers whose preprints consistently match your profile. these are people working on similar problems — filtered to exclude your existing co-authors.