Quality, completeness, and improvement queues
This screen shows the current corpus state without editorial promises or automatic publishing. It exposes coverage metrics, concrete issue groups, and routes to data that needs review.
Coverage scorecard
- Published dossiers
- 306
- Canonical sources
- 661
- RU coverage
- 96%
- KZ coverage
- 46%
- EN coverage
- 5%
- Dossiers with chapters
- 76%
- Event source coverage
- 14%
- Relationship source coverage
- 1%
- Corpus readiness
- 53%
- Evidence integrity
- 48%
- Monitoring dates
- 46%
- Source snapshots
- 38%
Metrics are calculated from the published database; they describe layer readiness, not a legal or editorial judgement of described actors.
Open issue groups
Each group is built from the live database and includes bounded examples so an editor, analyst, or developer can see what needs work.
Dossiers without RU version
Russian is the primary editorial version, so these profiles need priority preparation.
- open items
- 13
- samples
- 10
Dossiers without KZ version
Kazakh versions should appear after editorial review, not as machine-filled placeholders.
- open items
- 165
- samples
- 10
Dossiers without EN version
English versions support external readers and integrations, but are published only after explicit preparation.
- open items
- 290
- samples
- 10
Dossiers without structured biography
Biography facts power cards, comparisons, and brief summaries, so they should be separated from free text.
- open items
- 41
- samples
- 10
Dossiers without event chronology
Chronology is required for sequence checks, date filters, and analytical collections.
- open items
- 196
- samples
- 10
Dossiers without images
Images improve profile recognition, but must carry source, alt text, and rights metadata.
- open items
- 50
- samples
- 10
Dossiers without chapters
Chapters turn an index card into a real dossier and allow every content block to carry its own evidence state.
- open items
- 74
- samples
- 10
QazGeo location candidates awaiting review
Public maps should show accepted, reproducible locations rather than raw extraction candidates.
- open items
- 31
- samples
- 10
Mentions without publication dates
Dates are required for filters, timelines, deduplication, and monitoring freshness.
- open items
- 605
- samples
- 10
Sources without archival snapshots
Public URLs can disappear, so priority sources need snapshots, hashes, or archival URLs.
- open items
- 408
- samples
- 10
Documents without precise profile anchors
Documents become more useful when a profile links to a specific text block rather than the whole file.
- open items
- 1
- samples
- 1
Corpus enrichment queue
These workstreams are generated from live quality metrics. They show what is already connected to the database and what still needs editorial review.
Key relationship review
Relationships appear in the graph only after source linkage and editorial review; archival cross-references stay in the queue.
- Links
- 6 556
- Evidence
- 121
- Readiness
- 1%
Immutable source snapshots
Public URLs can change, so priority sources need snapshots or hashes.
- Snapshots
- 253
- Sources
- 661
- Coverage
- 38%
Reviewable statement layer
Draft facts are already structured, but the public statement layer opens only after review.
- Drafts
- 690
- Published
- 0
- With evidence
- 690
Monitoring dates and clusters
Mentions become useful when publication date, source, duplicate cluster and match quality are visible.
- Approved
- 1 119
- Dated
- 514
- Undated
- 605
QazGeo location review
Maps should expose accepted, reproducible locations rather than raw extraction candidates.
- Locations
- 31
- Candidates
- 31
- Reviewed
- 0
Documents and text blocks
Primary documents should be segmented into citable blocks and linked back to profiles.
- Documents
- 2
- Published
- 1
- Blocks
- 27
Analytical collection briefs
Collections should explain scope, inclusion criteria, source strength and unresolved questions.
- Dossiers
- 306
- Chapters
- 1 400
- Untranslated
- 468