Automated Newspaper Monitoring
Publisher-specific connectors will be enabled only after subscription access and monitoring rules are documented.
Not configuredIndexing Evaluation Lab
Indexing-first evaluation build for MDRC historical/backlog clippings and Filipiniana recent/current newspaper work. Abstracting is on hold and preserved only as a future-capable component.
Document / PDF Upload
Upload one clipping or a backlog batch. The original file is preserved while OCR/extracted text is retained as a searchable derivative with page mapping where available.
Manual Record Intake
Create a record when no usable digital source is available.
Evaluation Lab
Processor readiness
Backlog QA
Clipping quality screen
Screen uploaded clippings for missing metadata, unusually short text, and possible duplicates. Flags never delete records automatically.
MDRC historical indexing support
OCR correction queue
Shows OCR-backed records that still need page-level human review. Original OCR versions remain preserved.
Controlled vocabulary
Stage vocabulary
Load CSV, TSV, or JSON for structural validation and later Annif experiments. Import does not certify RL approval.
Staged vocabularies
Newspaper connector design
Non-live contract fixture
Exercise discovery metadata, human-triggered metadata promotion, and fail-closed retrieval boundaries without publisher credentials or live newspaper content.
Indexing Evaluation Lab
Development vs. gold sets
Gold sets can be frozen to prevent accidental edits after benchmark examples are agreed. Source/expectation manifests and processor-run manifests preserve reproducibility without automatically selecting a preferred processor.
MDRC full-text layer
Search searchable OCR / extracted text
Searches the current machine-readable derivative and returns the source page mapping when available. The original scan/PDF remains authoritative.
Work Queue