Compliance
Why this is safer
skai-science is a metadata-first watchtower. It indexes scientific publications by disease, stage, and bibliometric signal — but it does not host or republish publisher content. Below are the design choices that keep this product on the right side of copyright and terms of service.
Source choices
- PubMed / NCBI E-utilities: official public API, no PubMed logos used, NCBI disclaimer surfaced in admin. Rate ≤3 req/s without an API key.
- Europe PMC REST: full-text search + license-aware retrieval.
- bioRxiv / medRxiv: official JSON API and RSS. Authors retain copyright; license metadata respected per item.
- Crossref: citation counts, license URLs, funder data.
- OpenAlex: open work / source / author bibliometrics.
- ClinicalTrials.gov: phase 1 / phase 2 cross-link.
- Google Scholar: link-out only. We never scrape Scholar — that violates Google's terms — and we never frame or cache its content.
Per-item rights flags
Every Layer A row carries a SPDX license identifier and two boolean enforcement flags:can_display_abstract andcan_full_text_mine. The Catalog service consults these flags on every read, and the projector clears the abstract excerpt for any row that is not safe to display. A daily cron revalidates licenses for items under two years old; OA can revert, and we want to know.
Journal Impact Factor
The default ranking metric is CiteScore (open). Clarivate's Journal Impact Factor is only displayed when an admin attests to a Journal Citation Reports license — it is not enabled by default and is never written without that attestation.
No PDFs, no figures, no logos
We do not mirror publisher PDFs or images. We do not display NCBI, PubMed, Elsevier, Clarivate, or Google trademarks or logos. The "Open" link on every result navigates to the source URL — there is no in-app reader.
Takedown workflow
Anyone can submit a takedown or correction request via the takedown form. The request is written to a hash-chained audit log, and the affected row is moved into a curator review queue.