Where all of it came from

Sources

Every dataset here was published by someone else first. This is the full list of who, and how much of the catalogue each one accounts for.

  • 259 publishers
  • 4,748 datasets
  • 12,453,546 features

452 datasets carry no source URL (634,427 features). They are counted here rather than hidden — a sources page that quietly dropped them would overstate the very thing it exists to show.

Source material keeps its own terms, which are sometimes more restrictive than ours. What we derived is CC0; every download carries its source, licence and retrieval date inside the file.