Skip to content

Dataset download

Coming soon

A direct download of the Canadian Political Data dataset is not yet available. We're working through the privacy and data-scrubbing review needed to publish a snapshot that excludes user accounts, saved searches, billing data, and other non-public artifacts.

This page will be updated with the download link, full schema notes, and load instructions when the snapshot is published.

What you'll get when it ships

When the snapshot is published, this page will document:

  • A direct-download URL for the latest snapshot.
  • The dump format — currently planned as one or more gzipped PostgreSQL dumps (the same shape as sovpro db backup, i.e. pg_dump | gzip plain-SQL, restorable via gunzip -c | psql).
  • Step-by-step load instructions for Postgres 16 with PostGIS, pgvector, and unaccent.
  • The list of tables included (politicians, speeches with embeddings, bills, ridings, infrastructure scans) and the list excluded (user accounts, auth tokens, saved searches, reports, billing).

Refresh cadence (planned)

Once shipped, the dataset will be refreshed approximately weekly, after each significant ingest pass. The download URL will always point at the latest snapshot. If you need a frozen point-in-time copy, mirror the snapshot you used and cite the date you downloaded.

Licence (planned)

The published dataset will be released under a Canadian-friendly, non-commercial licence — most likely Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0), which is widely understood internationally and works under Canadian law. The exact terms will be locked in and published alongside the download.

In short: free for journalism, research, education, and civic-tech projects, with attribution. Commercial reuse requires a separate arrangement — contact us.

In the meantime: bootstrap your own copy

Until the snapshot is published, the working alternative is to run the ingestion pipeline locally and let it build the corpus from the upstream sources we ourselves ingest from. This is exactly what the production system does — just on your hardware.

The full how-to is on the Local installation page, specifically the Bootstrapping the dataset section.

In one paragraph: you bring up the Docker Compose stack, run the seed and ingest commands (federal MPs, provincial rosters, then per- jurisdiction Hansard), and the scanner + embedding service do the rest on a continuous loop. Building a complete current-session corpus from cold takes hours; building a complete historical corpus across every covered jurisdiction takes considerably longer — but you can scope to just the jurisdictions you care about and skip the rest.

This route also gives you continuously fresh data instead of weekly snapshots, which matters if you're tracking ongoing legislative activity in real time.

Want early access?

If you have a research, journalism, or civic-tech use case that would benefit from an early snapshot — even one with rougher edges than the public version will have — contact us. We're happy to share what we have under a short reuse agreement while the public version comes together.