270 lines
10 KiB
YAML
270 lines
10 KiB
YAML
original_entry:
|
|
name: USP/UNICAMP/UNESP
|
|
institution_type: EDUCATION_PROVIDER
|
|
source: CH-Annotator (latin_american_institutions_AUTHORITATIVE_ch_annotator.yaml)
|
|
identifiers:
|
|
- identifier_scheme: GHCID
|
|
identifier_value: BR-SP-XXX-E-U
|
|
- identifier_scheme: GHCID_NUMERIC
|
|
identifier_value: '2268910688007071813'
|
|
- &id001
|
|
identifier_scheme: OLD_ID
|
|
identifier_value: https://w3id.org/heritage/custodian/br/usp-unicamp-unesp
|
|
- identifier_scheme: GHCID_UUID
|
|
identifier_value: 9aca17ca-26ee-55a5-b0a0-097c4fa12dbf
|
|
identifier_url: urn:uuid:9aca17ca-26ee-55a5-b0a0-097c4fa12dbf
|
|
- identifier_scheme: GHCID_UUID_SHA256
|
|
identifier_value: 1f7cca3f-e1bf-8845-9ffd-e0c495e43ab4
|
|
identifier_url: urn:uuid:1f7cca3f-e1bf-8845-9ffd-e0c495e43ab4
|
|
- identifier_scheme: RECORD_ID
|
|
identifier_value: 019a58fd-3227-7644-9eff-6acf958d2228
|
|
identifier_url: urn:uuid:019a58fd-3227-7644-9eff-6acf958d2228
|
|
- &id002
|
|
identifier_scheme: OLD_ID
|
|
identifier_value: 2268910688007071813
|
|
locations:
|
|
- country: BR
|
|
region: SÃO PAULO
|
|
processing_timestamp: '2025-12-06T23:39:01.761474+00:00'
|
|
ghcid:
|
|
ghcid_current: BR-SP-LSD-E-UUU-uspunicampunesp
|
|
ghcid_original: BR-SA-XXX-E-UUU-uspunicampunesp
|
|
ghcid_uuid: 86bf913c-1f0c-59b9-92e5-19e3c5dec4a4
|
|
ghcid_uuid_sha256: 60bab83c-3643-8e83-2d8a-25b7117fb0e1
|
|
ghcid_numeric: 6970085942057234051
|
|
record_id: ed4e3e58-5021-42c8-9c5f-5ff71158093b
|
|
generation_timestamp: '2025-12-06T23:39:01.761474+00:00'
|
|
location_resolution:
|
|
method: GOOGLE_PLACES_GEONAMES
|
|
country_code: BR
|
|
region_code: SP
|
|
region_name: São Paulo
|
|
city_code: LSD
|
|
city_name: Loteamento Solar de Campinas
|
|
geonames_id: 13005473
|
|
feature_code: PPL
|
|
google_place_id: ChIJtU3SBbDGyJQRltcEzw5128Y
|
|
latitude: -22.818439299999998
|
|
longitude: -47.0647206
|
|
resolution_date: '2025-12-07T16:39:11.584916+00:00'
|
|
ghcid_history:
|
|
- ghcid: BR-SA-XXX-E-UUU-uspunicampunesp
|
|
ghcid_numeric: 6970085942057234051
|
|
valid_from: '2025-12-06T23:39:01.761474+00:00'
|
|
reason: Initial GHCID from CH-Annotator (latin_american_institutions_AUTHORITATIVE_ch_annotator.yaml)
|
|
valid_to: '2025-12-07T16:39:11.584916+00:00'
|
|
- ghcid: BR-SP-LSD-E-UUU-uspunicampunesp
|
|
ghcid_numeric: 6970085942057234051
|
|
valid_from: '2025-12-07T16:39:11.584916+00:00'
|
|
reason: 'Location resolved via Google Places + GeoNames: Loteamento Solar de Campinas (SP)'
|
|
custodian_name:
|
|
claim_type: custodian_name
|
|
claim_value: USP/UNICAMP/UNESP
|
|
source_type: ch_annotator
|
|
identifiers:
|
|
- identifier_scheme: GHCID
|
|
identifier_value: BR-SP-LSD-E-UUU-uspunicampunesp
|
|
- identifier_scheme: GHCID_UUID
|
|
identifier_value: 86bf913c-1f0c-59b9-92e5-19e3c5dec4a4
|
|
- identifier_scheme: GHCID_UUID_SHA256
|
|
identifier_value: 60bab83c-3643-8e83-2d8a-25b7117fb0e1
|
|
- identifier_scheme: GHCID_NUMERIC
|
|
identifier_value: '6970085942057234051'
|
|
- identifier_scheme: RECORD_ID
|
|
identifier_value: ed4e3e58-5021-42c8-9c5f-5ff71158093b
|
|
- *id001
|
|
- *id002
|
|
provenance:
|
|
data_source: CONVERSATION_NLP
|
|
data_tier: TIER_4_INFERRED
|
|
extraction_date: '2025-11-06T08:25:13.555410+00:00'
|
|
extraction_method: 'Created from CH-Annotator file: latin_american_institutions_AUTHORITATIVE_ch_annotator.yaml'
|
|
confidence_score: 0.7
|
|
notes:
|
|
- Canonical location added via normalize_custodian_files.py on 2025-12-08T23:10:53Z
|
|
- Canonical location added via normalize_custodian_files.py on 2025-12-08T23:46:58Z
|
|
- Canonical location added via normalize_custodian_files.py on 2025-12-09T06:48:23Z
|
|
- 'YouTube/Google Maps enrichment 2025-12-09T08:52:10Z: YouTube: not found'
|
|
schema_version: 2.0.0
|
|
enrichment_provenance:
|
|
wikidata_enrichment:
|
|
content_hash: sha256-3yTCklWNcI1ppijKW+E/DJfA4PeBTHwnAA934Y07sc8=
|
|
verified_at: '2025-12-28T19:57:38.147557+00:00'
|
|
wasDerivedFrom: https://www.wikidata.org/wiki/Q835960
|
|
google_maps_enrichment:
|
|
content_hash: sha256-hQO+DjkrA1HJBDGfx9dCKL/SsMoX4CpPn3wAa+zr/Sc=
|
|
verified_at: '2025-12-28T19:57:38.147580+00:00'
|
|
provenance_schema_version: '2.0'
|
|
standards_compliance:
|
|
- W3C PROV-O
|
|
- W3C SRI (content hashes)
|
|
generated_at: '2025-12-28T19:57:38.147590+00:00'
|
|
ch_annotator:
|
|
convention_id: ch_annotator-v1_7_0
|
|
convention_version: 1.7.0
|
|
entity_classification:
|
|
hypernym: GRP
|
|
hypernym_label: GROUP
|
|
subtype: GRP.EDU
|
|
subtype_label: EDUCATION_PROVIDER
|
|
ontology_class: schema:EducationalOrganization
|
|
alternative_classes:
|
|
- org:FormalOrganization
|
|
- rov:RegisteredOrganization
|
|
- glam:HeritageCustodian
|
|
extraction_provenance:
|
|
namespace: glam
|
|
path: /conversations/2025-09-22T14-40-15-0102c00a-4c0a-4488-bdca-5dd9fb94c9c5
|
|
timestamp: '2025-11-06T08:25:13.555410+00:00'
|
|
agent: claude-conversation
|
|
context_convention: ch_annotator-v1_7_0
|
|
annotation_provenance:
|
|
annotation_agent: opencode-claude-sonnet-4
|
|
annotation_date: '2025-12-06T21:13:56.173868+00:00'
|
|
annotation_method: retroactive CH-Annotator application via batch script
|
|
source_file: latin_american_institutions_AUTHORITATIVE.yaml
|
|
annotation_metadata:
|
|
confidence_score: 0.7
|
|
verified: false
|
|
verification_date:
|
|
verified_by:
|
|
entity_claims:
|
|
- claim_type: full_name
|
|
claim_value: USP/UNICAMP/UNESP
|
|
property_uri: skos:prefLabel
|
|
provenance:
|
|
namespace: glam
|
|
path: /conversations/2025-09-22T14-40-15-0102c00a-4c0a-4488-bdca-5dd9fb94c9c5
|
|
timestamp: '2025-11-06T08:25:13.555410+00:00'
|
|
agent: claude-conversation
|
|
context_convention: ch_annotator-v1_7_0
|
|
confidence: 0.7
|
|
- claim_type: institution_type
|
|
claim_value: EDUCATION_PROVIDER
|
|
property_uri: rdf:type
|
|
provenance:
|
|
namespace: glam
|
|
path: /conversations/2025-09-22T14-40-15-0102c00a-4c0a-4488-bdca-5dd9fb94c9c5
|
|
timestamp: '2025-11-06T08:25:13.555410+00:00'
|
|
agent: claude-conversation
|
|
context_convention: ch_annotator-v1_7_0
|
|
confidence: 0.95
|
|
- claim_type: ghcid
|
|
claim_value: BR-SP-XXX-E-U
|
|
property_uri: glam:ghcid
|
|
provenance:
|
|
namespace: glam
|
|
path: /conversations/2025-09-22T14-40-15-0102c00a-4c0a-4488-bdca-5dd9fb94c9c5
|
|
timestamp: '2025-11-06T08:25:13.555410+00:00'
|
|
agent: claude-conversation
|
|
context_convention: ch_annotator-v1_7_0
|
|
confidence: 1.0
|
|
integration_note:
|
|
created_from: latin_american_institutions_AUTHORITATIVE_ch_annotator.yaml
|
|
creation_date: '2025-12-06T23:39:01.761474+00:00'
|
|
creation_method: create_custodian_from_ch_annotator.py
|
|
google_maps_enrichment:
|
|
place_id: ChIJtU3SBbDGyJQRltcEzw5128Y
|
|
formatted_address: Cidade Universitária Zeferino Vaz - Barão Geraldo, Campinas - SP, 13083-970, Brazil
|
|
website: http://www.unicamp.br/
|
|
latitude: -22.818439299999998
|
|
longitude: -47.0647206
|
|
enriched_at: '2025-12-07T16:39:11.584916+00:00'
|
|
source: Google Places API (New)
|
|
_provenance:
|
|
content_hash:
|
|
algorithm: sha256
|
|
value: sha256-hQO+DjkrA1HJBDGfx9dCKL/SsMoX4CpPn3wAa+zr/Sc=
|
|
scope: enrichment_section
|
|
computed_at: '2025-12-28T19:57:38.147580+00:00'
|
|
prov:
|
|
wasDerivedFrom: https://maps.googleapis.com/maps/api/place/details/json?place_id=ChIJtU3SBbDGyJQRltcEzw5128Y
|
|
wasGeneratedBy:
|
|
'@type': prov:Activity
|
|
name: google_maps_api_fetch
|
|
used: https://maps.googleapis.com/maps/api/place
|
|
verification:
|
|
status: verified
|
|
last_verified: '2025-12-28T19:57:38.147584+00:00'
|
|
location:
|
|
latitude: -22.818439299999998
|
|
longitude: -47.0647206
|
|
coordinate_provenance:
|
|
source_type: GHCID_RESOLUTION_DIRECT
|
|
source_path: ghcid.location_resolution
|
|
entity_id: 13005473
|
|
resolution_method: GOOGLE_PLACES_GEONAMES
|
|
city: Loteamento Solar de Campinas
|
|
region: São Paulo
|
|
region_code: SP
|
|
country: BR
|
|
formatted_address: Cidade Universitária Zeferino Vaz - Barão Geraldo, Campinas - SP, 13083-970, Brazil
|
|
geonames_id: 13005473
|
|
feature_code: PPL
|
|
normalization_timestamp: '2025-12-09T06:48:23.379588+00:00'
|
|
youtube_status: NOT_FOUND
|
|
youtube_search_query: USP/UNICAMP/UNESP official
|
|
youtube_search_timestamp: '2025-12-09T08:52:10.887069+00:00'
|
|
wikidata_enrichment:
|
|
wikidata_id: Q835960
|
|
wikidata_url: https://www.wikidata.org/wiki/Q835960
|
|
label: University of São Paulo
|
|
description: public state university in Brazil
|
|
enrichment_timestamp: '2025-01-13T00:00:00Z'
|
|
enrichment_method: manual_wikidata_lookup
|
|
identifiers:
|
|
VIAF: '138319002'
|
|
GND: 1010570-0
|
|
LCNAF: n79143148
|
|
official_website: https://www.usp.br
|
|
notes: USP/UNICAMP/UNESP digital consortium repository - using USP as primary Wikidata entry
|
|
_provenance:
|
|
content_hash:
|
|
algorithm: sha256
|
|
value: sha256-3yTCklWNcI1ppijKW+E/DJfA4PeBTHwnAA934Y07sc8=
|
|
scope: enrichment_section
|
|
computed_at: '2025-12-28T19:57:38.147557+00:00'
|
|
prov:
|
|
wasGeneratedBy:
|
|
'@type': prov:Activity
|
|
name: wikidata_api_fetch
|
|
used: https://www.wikidata.org/w/rest.php/wikibase/v1
|
|
wasDerivedFrom: https://www.wikidata.org/wiki/Q835960
|
|
generatedAtTime: '2025-01-13T00:00:00Z'
|
|
verification:
|
|
status: verified
|
|
last_verified: '2025-12-28T19:57:38.147562+00:00'
|
|
wikidata_entity_id: Q835960
|
|
logo_enrichment:
|
|
enrichment_timestamp: '2025-12-22T14:51:37.122002+00:00'
|
|
source_url: http://www.unicamp.br
|
|
extraction_method: crawl4ai
|
|
claims:
|
|
- claim_type: logo_url
|
|
claim_value: https://unicamp.br/wp-content/themes/bx-unicamp-multisite/assets/img/logo-unicamp.svg
|
|
source_url: http://www.unicamp.br
|
|
css_selector: '#body > header.header.desktop-header > div.info-wrapper:nth-of-type(2) > div.container > div.row.d-flex > div.col-auto.d-flex > div.site-logos > div.main-logo.light-logo > a > img'
|
|
retrieved_on: '2025-12-22T14:51:37.122002+00:00'
|
|
extraction_method: crawl4ai_header_logo
|
|
detection_confidence: high
|
|
alt_text: Site da Unicamp
|
|
- claim_type: favicon_url
|
|
claim_value: https://unicamp.br/wp-content/uploads/sites/33/2024/03/cropped-logo_unicamp_512-180x180.png
|
|
source_url: http://www.unicamp.br
|
|
css_selector: '[document] > html > head > link:nth-of-type(22)'
|
|
retrieved_on: '2025-12-22T14:51:37.122002+00:00'
|
|
extraction_method: crawl4ai_link_rel
|
|
favicon_type: ''
|
|
favicon_sizes: ''
|
|
- claim_type: og_image_url
|
|
claim_value: https://www.unicamp.br/wp-content/uploads/sites/33/2023/12/atu_reitoria_20200804_capa.jpg
|
|
source_url: http://www.unicamp.br
|
|
css_selector: '[document] > html > head > meta:nth-of-type(10)'
|
|
retrieved_on: '2025-12-22T14:51:37.122002+00:00'
|
|
extraction_method: crawl4ai_meta_og
|
|
summary:
|
|
total_claims: 3
|
|
has_primary_logo: true
|
|
has_favicon: true
|
|
has_og_image: true
|
|
favicon_count: 3
|