Corpus & data pipeline
Backfill
A backfill retroactively fills in missing or newly added fields across records that already exist, rather than only handling new ones.
A backfill applies a new or improved process to the existing corpus so older records catch up. When a pipeline adds a field, a backfill computes it for everything already stored - not just future intake.
Backfills are usually batch jobs, run once to close a gap, then folded into the normal pipeline going forward.
In GraphCanon
When GraphCanon adds a signal (say, compatibility edges or decision facts), a scoped backfill run populates it across the corpus, with progress logged per run.
Related terms
Last reviewed 2026-07-09