glam/enrich_brazil_batch13.py

#!/usr/bin/env python3
"""
Brazil Batch 13 Wikidata Enrichment Script

Target: 10-12 institutions to reach ~60-65% coverage (73-75 out of 121)
Focus: National institutions, state museums, major universities

This script uses manual Wikidata searches and documents Q-numbers found.
Run the actual enrichment by calling wikidata-authenticated_search_entity for each institution.
"""

import yaml
from datetime import datetime, timezone
from pathlib import Path

print("="*80)
print("BRAZIL BATCH 13 WIKIDATA ENRICHMENT")
print("="*80)
print("Target: 10-12 institutions (aiming for 60-65% coverage)")
print("="*80)
print()

# =============================================================================
# BATCH 13 TARGET INSTITUTIONS
# =============================================================================

BATCH13_TARGETS = [
    {
        'name': 'Instituto Moreira Salles',
        'id': 'https://w3id.org/heritage/custodian/br/instituto-moreira-salles',
        'search_query': 'Instituto Moreira Salles Brazil',
        'context': 'Major cultural institute with photographic collections',
        'expected_qid': 'Q10302915',  # Known from research
        'priority': 1
    },
    {
        'name': 'Sistema Brasileiro de Museus',
        'id': 'https://w3id.org/heritage/custodian/br/sistema-brasileiro-de-museus-sbm',
        'search_query': 'Sistema Brasileiro de Museus',
        'context': 'National museum system created 2004, coordinated by IBRAM',
        'priority': 1
    },
    {
        'name': 'Brasiliana Fotográfica',
        'id': 'https://w3id.org/heritage/custodian/br/brasiliana-fotografica',
        'search_query': 'Brasiliana Fotográfica',
        'context': 'Inter-institutional photography collaboration',
        'priority': 1
    },
    {
        'name': 'Universidade Federal de Rondônia',
        'id': '3008281717687280329',
        'search_query': 'Universidade Federal de Rondônia',
        'context': 'Federal university in Rondônia (UNIR)',
        'expected_qid': 'Q10365614',  # Federal universities follow pattern
        'priority': 1
    },
    {
        'name': 'Fundação de Cultura Elias Mansour',
        'id': 'https://w3id.org/heritage/custodian/br/ac-funda-o-de-cultura-elias-mansour-fem',
        'search_query': 'Fundação de Cultura Elias Mansour Acre',
        'context': 'State cultural foundation in Acre',
        'priority': 2
    },
    {
        'name': 'Museu dos Povos Acreanos',
        'id': 'https://w3id.org/heritage/custodian/br/ac-museu-dos-povos-acreanos',
        'search_query': 'Museu dos Povos Acreanos Rio Branco',
        'context': 'Museum in Rio Branco (opened 2023)',
        'priority': 2
    },
    {
        'name': 'Museu Histórico de Alcântara',
        'id': 'https://w3id.org/heritage/custodian/br/mt-museu-hist-rico',
        'search_query': 'Museu Histórico de Alcântara Maranhão',
        'context': 'Historical museum in Alcântara, Maranhão',
        'priority': 2
    },
    {
        'name': 'Secretaria de Estado da Cultura do Amapá',
        'id': '1423599463777727402',
        'search_query': 'Secretaria Cultura Amapá',
        'context': 'State culture secretariat of Amapá',
        'priority': 3
    },
    {
        'name': 'Secretaria de Estado da Cultura do Tocantins',
        'id': '709508309148680086',
        'search_query': 'Secretaria Cultura Tocantins',
        'context': 'State culture secretariat of Tocantins',
        'priority': 3
    },
    {
        'name': 'Instituto Histórico e Geográfico de Alagoas',
        'id': '2519599505258789521',
        'search_query': 'Instituto Histórico Geográfico Alagoas',
        'context': 'Historical and geographic institute in Alagoas',
        'priority': 3
    },
    {
        'name': 'Sistema de Museus de Ouro Preto',
        'id': 'https://w3id.org/heritage/custodian/br/mg-ouro-preto-system',
        'search_query': 'Sistema Museus Ouro Preto',
        'context': "Brazil's first municipal museum network (2006)",
        'priority': 3
    },
    {
        'name': 'Museu Histórico de Goiás',
        'id': 'https://w3id.org/heritage/custodian/br/go-museu-hist-rico-mham',
        'search_query': 'Museu Histórico de Goiás',
        'context': 'Historical museum in Goiás state',
        'priority': 3
    }
]

print("BATCH 13 TARGET INSTITUTIONS")
print("="*80)
print(f"Total targets: {len(BATCH13_TARGETS)}")
print()

for idx, target in enumerate(BATCH13_TARGETS, 1):
    print(f"{idx}. {target['name']}")
    print(f"   Priority: {target['priority']}")
    print(f"   Search query: {target['search_query']}")
    print(f"   Context: {target['context']}")
    if 'expected_qid' in target:
        print(f"   Expected Q-ID: {target['expected_qid']}")
    print()

print("="*80)
print("INSTRUCTIONS FOR ENRICHMENT:")
print("="*80)
print("""
1. For each institution above, call the wikidata-authenticated_search_entity tool
   with the search query provided

2. Verify the Q-number returned matches the institution by checking:
   - Label matches the institution name
   - Description mentions correct location (Brazil, state)
   - Instance of (P31) is correct type (museum, university, etc.)

3. Record results in BATCH13_MATCHES dictionary below

4. For institutions not found via search, use wikidata-authenticated_execute_sparql
   with geographic and type filters (P17=Q155 for Brazil, P131 for state)

5. After collecting all Q-numbers, create batch13_enriched.yaml with verified matches

6. Run merge script to integrate into main dataset
""")

print("\nReady to begin searches!")
print("Use the wikidata-authenticated_search_entity tool for each target above.")
print("="*80)

# Template for recording results
BATCH13_MATCHES = {
    # Will be filled in as searches are completed
    # Example format:
    # "Instituto Moreira Salles": {
    #     "qid": "Q10302915",
    #     "label": "Instituto Moreira Salles",
    #     "confidence": 0.95,
    #     "notes": "Verified via search - cultural institute"
    # }
}

# Calculate current status
current_with_wikidata = 67
total_brazilian = 121
current_coverage = (current_with_wikidata / total_brazilian) * 100

print(f"\nCURRENT STATUS:")
print(f"  Institutions with Q-numbers: {current_with_wikidata}/121 ({current_coverage:.1f}%)")
print(f"  Target after Batch 13: 73-77/121 (60-64%)")
print(f"  Need to find: 6-10 Q-numbers")
print()