v1.1.0 - reproduction snapshot

Data snapshot the current site is built from. Adds the country tags on harvested rows (the /in/ pages) and the repaired date_prefix fields that v1.0.0 predates, so scripts/reproduce.sh matches the published pages again.

sha256: baaa67da9dc22d017be162b065590c2ed35a246bfe71661c823b8e26e8360c4e

v1.0.0 — reproduction snapshot

The data snapshot the published site was built from.

slashyear-data-snapshot.tar.gz (105 MB) contains data/claims/ — the extracted, cleaned, classified sentences — plus data/raw/ (the Wikipedia revisions they were extracted from, so verify.py can re-check every citation offline) and the caches the remaining gates need.

scripts/bootstrap.sh    # download, checksum, unpack into ./data
scripts/reproduce.sh    # rebuild every page and run the gates; prints PASS or FAIL

sha256 0801b65e3cd905fd21e120e618aa6e9f2eb44b516c0d0396c3d90c94ce20bd07

To rebuild from Wikipedia itself instead of from this snapshot, start at pipeline/harvest.py and follow the stage order in docs/PIPELINE.md.

Rows are CC BY-SA 4.0; code is Apache-2.0.