IRCC Spousal Sponsorship Advanced Analytics: A 2026 Transparency Dataset and Comparative Review

Correction notice, version 1.2, 3 November 2026. No government value changed. Six cells were reworded, and each keeps its YES or NOT_PUBLICLY_ESTABLISHED answer. gba_plus, both rows: the words "the analysis itself is not public" went further than the source, so the cell now says the AIA (Algorithmic Impact Assessment) answers No on whether this is publicly available, and the overseas cell now says the IRCC notice lists the analysis. model_monitoring, inland row: the earlier wording did not name its source, so the cell now names the inland AIA and IRCC's December 2023 material. hidden_features_or_weights, both rows: now adds each AIA's answer on whether the algorithm is a trade secret (inland No, overseas Yes). peer_review_public_resource, overseas row: now calls the resource an executive summary and also names v1.2. In the JSON: the version was updated, control_issues (internal tracking numbers) was removed, peer_review_note was reworded, PARTIAL was defined in schema_note, professional_review was limited to the v0.2 interpretation, its boundary now also names v1.2, and the v1.1 notice moved to earlier_correction_notices. The README was updated to match, and the data file names now say v1.2. What did not change: the two records, the 28 fields and the other 50 of the 56 cells.

Machine-readable 2026 Canadian immigration transparency dataset comparing publicly disclosed IRCC advanced-analytics use in in-Canada and overseas spouse/partner processing. It normalizes official Canada.ca and Open Government records across 28 fields covering positive eligibility automation, human decision boundaries, triage, safeguards, oversight and disclosure gaps. Inland and overseas systems are treated separately, and NOT_PUBLICLY_ESTABLISHED is used where the reviewed public source set does not establish a proposition. The reviewed source set includes the AIA files attached to each Open Government record. NOT_PUBLICLY_ESTABLISHED does not mean No. The release contains no confidential client records, case outcomes or unpublished MAK case data.

Professional review: Omer Khalil, RCIC R710149, of the v0.2 interpretation from which v1.0 is derived. That review predates the 3 November 2026 corrections in v1.1 and v1.2. This is professional RCIC review, not academic peer review or Government of Canada endorsement.

History: correction notice published with version 1.1, 3 November 2026. It describes the change from 1.0 to 1.1, not this version. Version 1.0 marked three in-Canada (inland) points as NOT_PUBLICLY_ESTABLISHED: Gender Based Analysis Plus, quality assurance, and privacy and security design. That was wrong or incomplete. The Algorithmic Impact Assessment (AIA) attached to the Open Government record that the dataset cites answers these points. What changed in 1.1: those three cells are corrected. Both rows now say what the AIAs state about recourse and that the Gender Based Analysis Plus itself is not public. The 2026 context cell now says what the 2026 IRCC binder says, and separately that the tool is listed in the IRCC inventory dated 2025-11-17. The peer-review cell and the JSON notes now also refer to v1.1, and the JSON title, status and date fields were updated for this release. The overseas wording on impact level, officer route and privacy now follows its AIA, and the inland triage cell now notes what the inland AIA says. What did not change: the two records, the 28 fields, the finding that IRCC discloses automated positive eligibility processing for routine cases, and that admissibility and the final decision remain with an officer.