- Fixed _vector_search() to check uses_named_vectors() before adding 'using' parameter
- Fixed _person_vector_search() to detect person collection vector size and use appropriate model
- Resolves 'Not existing vector name error: openai_1536' for single-vector collections
- Resolves embedding dimension mismatch between heritage_custodians (1536-dim) and heritage_persons (384-dim)
- Institution Browser: multi-select for types and countries
- URL query param sync for shareable filter URLs
- New utility: countryNames.ts with flag emoji support
- New utility: imageProxy.ts for image URL handling
- New component: SearchableMultiSelect dropdown
- Career timeline CSS and component updates
- Media gallery improvements
- Lazy load error boundary component
- Version check utility
- Add digital platform discovery data with provenance
- Cleanup duplicate/incorrect custodian entries
- Add GHCID collision resolution suffixes where needed
- Update person entity profiles with career history
- Add GLAMORCUBESFIXPHDNT heritage type detection for person profiles
- Two-stage classification: blocklist non-heritage orgs, then match keywords
- Special handling for Digital (D) type: requires heritage org context
- Add career_history heritage_relevant and heritage_type fields
- Add exponential backoff retry for Anthropic API overload errors
- Fix DSPy 3.x async context with dspy.context() wrapper
- Add refresh button to assistant messages for re-running queries with fresh results
- Highlight refresh button (amber) for cached responses to draw attention
- Add spinning icon animation while refreshing
- Fix cache clear to return detailed success/failure status for local vs shared cache
- Add bypass cache toggle that forces fresh queries (one-shot, resets after query)
- Add Dutch/English translations for new UI elements
- Add SyncPanel component with bilingual (NL/EN) support
- Add relative URL handling for production (bronhouder.nl)
- Integrate SyncPanel into Database page
- Show sync status for all 4 databases (DuckLake, PostgreSQL, Oxigraph, Qdrant)
- Support dry-run mode and file limit options
- Implemented a new script `test_pico_arabic_waqf.py` to test the GLM annotator's ability to extract person observations from Arabic historical documents.
- The script includes environment variable handling for API token, structured prompts for the GLM API, and validation of extraction results.
- Added comprehensive logging for API responses, extraction results, and validation errors.
- Included a sample Arabic waqf text for testing purposes, following the PiCo ontology pattern.
- Download GeoNames JP postal code database (142K entries)
- Create geocode_japan_postal.py with postal code lookup
- Handle unicode hyphen variants in postal codes
- Add manual mappings for remote Tokyo islands (Hachijojima, Miyakejima)
- Implement prefix fallback for company postal codes
- Total JP files geocoded: 540 (99.81% coverage)
This brings overall geocoding coverage from 97.84% to 99.81%
- Improved city name cleaning:
- Roman numeral district suffixes (Kolín V. -> Kolín)
- City + country suffixes (Genève 4 - Suisse -> Genève)
- Czech postal notation (p. Luka nad Jihlavou -> Luka nad Jihlavou)
- Historical city names (Gottwaldov -> Zlín, renamed 1990)
- Manual mappings for Swiss districts (Lugano Massagno -> Lugano)
- Handle Czech address patterns:
- House numbers with čp./č.p. prefix
- X nad/pod Y town names (rivers/landmarks)
- Hyphenated district names (Město-Část)
- Trailing numbers and suffixes
- Improved city name normalization to handle:
- St. Gallen / St.Gallen -> Sankt Gallen
- Canton suffixes (Buchs SG, Brugg AG)
- Hyphenated districts (Bernex - Genève)
- Postal codes with slashes (Ecublens/VD)
- German prepositions (Hausen b. Brugg)
- Created scripts/geocode_from_city_name.py for unified geocoding
- Introduced LEGAL-FORM-FILTER rule to standardize CustodianName by removing legal form designations.
- Documented rationale, examples, and implementation guidelines for the filtering process.
docs: Create README for value standardization rules
- Established a comprehensive README outlining various value standardization rules applicable to Heritage Custodian classes.
- Categorized rules into Name Standardization, Geographic Standardization, Web Observation, and Schema Evolution.
feat: Implement transliteration standards for non-Latin scripts
- Added TRANSLIT-ISO rule to ensure GHCID abbreviations are generated from emic names using ISO standards for transliteration.
- Included detailed guidelines for various scripts and languages, along with implementation examples.
feat: Define XPath provenance rules for web observations
- Created XPATH-PROVENANCE rule mandating XPath pointers for claims extracted from web sources.
- Established a workflow for archiving websites and verifying claims against archived HTML.
chore: Update records lifecycle diagram
- Generated a new Mermaid diagram illustrating the records lifecycle for heritage custodians.
- Included phases for active records, inactive archives, and processed heritage collections with key relationships and classifications.
- Final 42 files updated
- Normalization complete: all 27,511 custodian files have location block
- 15,419 files have coordinates with coordinate_provenance
- 12,092 files have address-only location blocks
- 2,546 files updated with location blocks
- All 27,511 custodian files now have location: block
- 15,421 files have coordinates with coordinate_provenance
- 12,090 files have address-only location blocks
Key changes:
- Created scripts/lib/safe_yaml_update.py with PROTECTED_KEYS constant
- Fixed enrich_custodians_wikidata_full.py to re-read files before writing
(prevents race conditions where another script modified the file)
- Added safety check to abort if protected keys would be lost
- Protected keys include: location, original_entry, ghcid, provenance,
google_maps_enrichment, osm_enrichment, etc.
Root cause of data loss in 62fdd35321:
- Script loaded files into list, then processed them later
- If another script modified files between load and write, changes were lost
- Now files are re-read immediately before modification
Per AGENTS.md Rule 5: NEVER Delete Enriched Data - Additive Only
- Fixed 469 JP files missing location: blocks (had data in original_entry.locations)
- Fixed 117 additional JP files found in second pass
- 1 EG file skipped (no location source data available)
- Total files with location: blocks now 27,459 out of 27,511 (99.8%)
- Also includes YAML formatting standardization (line wrapping)
Recovery from data loss in commit 62fdd35321 is now complete.
