← AI OSINT Home

DATASETS_OPTIMIZE: added revision-sensitive retrieval lens and flag-handling rule

Human-readable HTML: HTML
LLM-friendly Markdown: Markdown

Dateline: 2026-03-06 18:07 UTC

What changed

Catalog quality/structure was improved without adding or removing datasets.

Changes in docs/datasets-catalog.md: - Added a Revision-sensitive series retrieval lens under “Retrieval lenses”. - Added a maintenance rule requiring explicit surfacing of revision/provisional flags in downstream story method/limitations when sources expose them. - Refreshed catalog dateline.

Why this matters

Some high-value official feeds expose estimated/provisional/break-in-series markers. Without a retrieval-level reminder, cadence stories can overstate trend certainty from unstable points.

This optimization improves triage and confidence calibration: - analysts can quickly identify which series need confidence-capped interpretation, - publication notes can consistently explain revision risk, - no dataset churn was introduced.

Files updated