Licence chosen by each author per item.
Sources and verification
Where GAKN’s knowledge comes from, what we may show from each source, and how we label trust.
Crossref
RespondingBibliographic metadata usable for any purpose; abstracts may remain under publisher copyright.
DOAJ
FailingMetadata CC0; articles open access under the journal licence.
Data.gov
RespondingUS federal data generally public domain; each dataset states its licence.
Europe PMC
RespondingArticle licence varies; abstracts excerpted only.
Hacker News
RespondingPosts belong to their authors; titles and links only.
Library of Congress
FailingRights vary per item.
OpenAlex
Needs a free key (owner action)Scholarly metadata CC0.
PeerTube (SepiaSearch index)
RespondingVideos hosted by independent PeerTube instances; licence set per video.
Public-domain books (in the US).
Wikidata
RespondingStructured data CC0 1.0.
Wikimedia Commons
RespondingLicence set per file.
Wikipedia
RespondingText CC BY-SA 4.0; commercial reuse allowed with attribution and share-alike.
Zenodo
RespondingRecord metadata CC0; files carry their own licence.
arXiv
RespondingMetadata CC0; each paper carries its own licence.
How verification works
- Every result names its source and links to the original. GAKN does not claim ownership of material from other sources.
- Authors attach claims to their work. Anyone signed in can add evidence that supports or contradicts a claim, with a link and a short quote.
- A claim with only supporting evidence is “Supported”, with only contradicting evidence “Contradicted”, with both “Disputed”. An item with any disputed or contradicted claim is labelled “Disputed”.
- Publishers declare how a text was made: by a person, with AI assistance, generated by AI, or machine-translated. Machine translations always link to the original.
Not connected, and why
- YouTube
- YouTube’s developer policies forbid offering an API that gives access to YouTube data, selling access, and caching beyond 30 days. GAKN’s agent API would conflict, so YouTube is not connected.
- Internet Archive
- Its terms historically limit use to scholarship and research purposes; pending the owner’s review of the current terms.
- Open Library
- Its guidelines say the APIs are not meant as a data backend for third-party services; bulk monthly dumps are the permitted route.
- Semantic Scholar
- Needs an API key and a licence review (parts of the data are non-commercial).
- Europeana · Stack Exchange · Openverse
- Free API keys are required; they will be connected once the owner registers them.
- Reddit · X · LinkedIn · Facebook
- Closed platforms without permitted bulk access for this use.