The JSON is keyed to internal IDs as opposed to IRIs, which...

@epoz @datasetteproj The JSON is keyed to internal IDs as opposed to IRIs, which are teensy.

(With NT I replaced the namespaces with their well known prefixes and that got me down to like 200gb.)

The JSON was grisly at first but I got there with lbzcat piping to parallel piping to jq, will share.