📄 PressMint QuickCheck: Operationalising Readiness Diagnostics for Interoperable Historical Newspaper Corpora

Summary

Digitised historical newspapers usually arrive as mixed library exports, such as PDF with embedded OCR, OCR text folders, ALTO/METS-ALTO, partial TEI or IIIF manifests. This makes it hard to judge early whether a collection is fit for the PressMint interoperable-corpus workflow. The authors present PressMint QuickCheck, a Colab notebook that needs no installation. It ingests a ZIP package or a folder of IIIF manifests and works at the level of individual issues. It runs only the checks that match the components it detects: inventory, metadata completeness (ISO 8601 dates, stable IDs, rights, provenance, BCP 47 language tags), OCR triage signals and IIIF Presentation API 3.0 validity. It outputs an HTML readiness report, a minimal CSV manifest and a JSON scorecard, with P0/P1/P2 priorities linked to remediation actions. Declared content and metadata languages are treated as core interoperability signals, given the multilingual nature of PressMint and ParlaMint. A demo on a PDF+OCR sample and an IIIF sample yields one unified manifest. The paper reports no quantitative evaluation. Its value lies in making onboarding transparent and comparable for small DH teams and libraries.

Key points

• Three artefacts are produced: an HTML readiness report, a minimal normalised manifest (CSV, one row per issue) and a scorecard (JSON with an optional per-check CSV). The scorecard gives a suitability_score from 0 to 100 and a suitability_level from 0 to 4, where 0 means unusable and 4 means "PressMint-ready". Weights are tentative v1 and configurable.
• Priorities follow the most severe flag. P0 means no usable text or structure, which blocks onboarding. P1 covers fixable gaps such as a missing language declaration, an unparseable date, missing rights or provenance, empty or low-signal OCR, or invalid IIIF. P2 covers non-blocking issues, e.g. P2:generated_id.
• Language fields (content_language, metadata_language, language_present) are always reported. If no language declaration is present, a P1 flag is raised.
• Version 2.0 adds P2:fields_from_defaults, so that values filled in from notebook defaults are no longer mistaken for metadata declared by the provider. This gap was found during external testing.
• The scope is deliberately limited: no TEI conversion, article segmentation or automatic language identification, and no measurement of OCR accuracy. The notebook is available in English and Spanish at github.com/ebattanermoro/PressMint-Quickcheck.