Open data infrastructure gaps are the specific datasets and publication failures that block accountability analysis even when other data is available. This page catalogs what’s missing across the federal and Alberta records, and what each gap blocks.
I
55 dataset classes that bear on this chapter, federal vs Alberta side-by-side
55 dataset classes · Federal vs Alberta
Federal
Alberta
Legend bulk machine-readable available · no bulk export gated · account required partial coverage not published· click any row for license, lag, ID, source
II
The full catalog follows below. These are the gaps that recur across the most other chapters.
01 · People
Bulk access to incorporation date, directors, and addresses is the single biggest blocker on the site — it gates Ghost Capacity, Zombie Recipients, and Related Parties from making canonical claims.
02 · Procurement & Contracts
There is no shared identifier that survives across grants, contracts, and disclosures. Without it, every cross-dataset analysis depends on name-matching heuristics that can never be enforcement-grade.
03 · Procurement & Contracts
Alberta publishes the sole-source flag but not the recorded justification. The justification text is the single field that turns a flag into a defensible audit signal.
04 · Outcomes & Reconciliation
Estimates and disclosures live in different formats with no shared keys, so "did we spend what we budgeted?" can't be answered without manual stitching.
05 · Infrastructure
There is no public scorecard for whether each dataset is published as CSV, API, PDF, or scrape-only — so format quality silently slides without consequence.
III
From the live ab_spending data-quality log and the active spec roadmap. Grouped by category; each entry is a real gap, not an aspiration.
IV
Ten representative AB dataset classes graded by what's actually published, not what should be. The grade reflects ingestion friction, not data quality.
A = stable open API with versioning. B = CSV / structured download with content-hash tracking. C = HTML table scrape required. D = PDF-only, parser-dependent. F = no published form (crawler required).
| Dataset class | Current format | Grade | Note |
|---|---|---|---|
| Blue Book — supplies & procurement | CKAN XLSX (annual) | B | Stable schema, indexed on CKAN. No API yet. |
| Historical government estimates | CKAN PDF (2000–2018) | D | Three layout eras across 20 years. Spec 017 parser closes the gap; the upstream is unchanged. |
| Ministers' office expenses | CKAN PDF (~125 packages) | D | Per-ministry monthly; attribution lives only in filenames. |
| Lobbyist registry | Live Oracle APEX portal | F | No bulk export. 19,453 records behind paginated search. |
| Political contributions (Elections AB) | Annual XLSX/PDF | B | Threshold-filtered. Amendment history needs dedup. |
| PSBCTA executive compensation | Consolidated CSV | B | Single endpoint. ~400 body names need aliasing. |
| Provincial royalty revenue | CKAN XLSX (1970+) | B | Long history. 8 unmapped category aggregates. |
| Education property tax requisition | CKAN PDF (annual, 2019+) | D | Per-municipality page layout. Limited history. |
| Provincial debt (TRM quarterly) | CKAN PDF (quarterly) | D | Layout drift across fiscal years. Point-in-time, not FY-aligned. |
| School authority funding | CKAN PDF per authority (1,296 files) | D | 1 PDF per authority per year. pdfplumber extracts cleanly; grain is the problem. |