Correct care mixed identifiers sidecar offsets

corpus/care/care_mixed_identifiers_seed_001.gold.json

@@ -1,35 +1 @@
-{
-  "document_id": "care_mixed_identifiers_seed_001",
-  "domain": "care",
-  "document_type": "synthetic_care_mixed_identifiers_note",
-  "language": "nl",
-  "source_file": "corpus/care/care_mixed_identifiers_seed_001.txt",
-  "synthetic": true,
-  "labels": [
-    {"id":"L001","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"C-118","start":6,"end":11,"sensitivity":"care_location_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Room reference can be identifying in care context."},
-    {"id":"L002","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"Rozenhof 3","start":28,"end":38,"sensitivity":"care_location_reference","required":true,"acceptable_entity_types":["LOCATION","ORGANIZATION","NL_HEALTHCARE_REFERENCE"],"notes":"Department reference can be identifying in care context."},
-    {"id":"L003","entity_class":"INCIDENT_NUMBER","text":"MIC-2026-7720","start":56,"end":69,"sensitivity":"incident_reference","required":true,"acceptable_entity_types":["NL_INCIDENT_NUMBER","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care incident reference."},
-    {"id":"L004","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"ZORG-CL-2026-88990","start":94,"end":112,"sensitivity":"care_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care client reference."},
-    {"id":"L005","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"ZD-2026-55661","start":128,"end":141,"sensitivity":"care_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care dossier reference."},
-    {"id":"L006","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"MED-2026-AX91","start":157,"end":170,"sensitivity":"medication_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Medication/admin code."},
-    {"id":"L007","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"VOS-UNIT-2026-441","start":183,"end":200,"sensitivity":"device_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care device reference."},
-    {"id":"L008","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"11223344556","start":235,"end":246,"sensitivity":"professional_reference","required":true,"acceptable_entity_types":["NL_BIG_NUMBER","NL_HEALTHCARE_REFERENCE"],"notes":"BIG-number-like value."},
-    {"id":"L009","entity_class":"PERSON","text":"Hassan El Amrani","start":295,"end":311,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PERSON"],"notes":"Person name."},
-    {"id":"L010","entity_class":"PERSON","text":"Mila van Dijk","start":315,"end":328,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PERSON"],"notes":"Person name."},
-    {"id":"L011","entity_class":"EMAIL","text":"hassan.elamrani@example.test","start":349,"end":377,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["EMAIL_ADDRESS"],"notes":"Reserved synthetic email domain."},
-    {"id":"L012","entity_class":"PHONE","text":"06 55667788","start":381,"end":392,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PHONE_NUMBER","NL_PHONE_NUMBER"],"notes":"Synthetic mobile number."}
-  ],
-  "preserve_terms": [
-    {"term":"Kamer","start":0,"end":5,"reason":"care location context term"},
-    {"term":"afdeling","start":19,"end":27,"reason":"care location context term"},
-    {"term":"cliënt","start":397,"end":403,"reason":"care role/context term"},
-    {"term":"arts","start":259,"end":263,"reason":"care role/context term"},
-    {"term":"verpleegkundige","start":415,"end":430,"reason":"care role/context term"},
-    {"term":"zorgmedewerker","start":437,"end":451,"reason":"care role/context term"}
-  ],
-  "known_traps": [
-    {"text":"Kamer C-118","trap_type":"care_location_context_needs_review","notes":"Room references may be identifying in care context and need explicit handling."},
-    {"text":"afdeling Rozenhof 3","trap_type":"care_location_context_needs_review","notes":"Department references may be identifying depending on context."},
-    {"text":"cliënt","trap_type":"care_role_word_must_not_be_person","notes":"Generic care role word should remain readable."}
-  ]
-}
+{"document_id":"care_mixed_identifiers_seed_001","domain":"care","document_type":"synthetic_care_mixed_identifiers_note","language":"nl","source_file":"corpus/care/care_mixed_identifiers_seed_001.txt","synthetic":true,"labels":[{"id":"L001","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"C-118","start":6,"end":11,"sensitivity":"care_location_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Room reference can be identifying in care context."},{"id":"L002","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"Rozenhof 3","start":29,"end":39,"sensitivity":"care_location_reference","required":true,"acceptable_entity_types":["LOCATION","ORGANIZATION","NL_HEALTHCARE_REFERENCE"],"notes":"Department reference can be identifying in care context."},{"id":"L003","entity_class":"INCIDENT_NUMBER","text":"MIC-2026-7720","start":56,"end":69,"sensitivity":"incident_reference","required":true,"acceptable_entity_types":["NL_INCIDENT_NUMBER","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care incident reference."},{"id":"L004","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"ZORG-CL-2026-88990","start":93,"end":111,"sensitivity":"care_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care client reference."},{"id":"L005","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"ZD-2026-55661","start":127,"end":140,"sensitivity":"care_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care dossier reference."},{"id":"L006","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"MED-2026-AX91","start":156,"end":169,"sensitivity":"medication_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Medication/admin code."},{"id":"L007","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"VOS-UNIT-2026-441","start":182,"end":199,"sensitivity":"device_reference","required":true,"acceptable_entity_types":["NL_HEALTHCARE_REFERENCE","NL_SUSPICIOUS_REFERENCE_CANDIDATE"],"notes":"Care device reference."},{"id":"L008","entity_class":"MEDICAL_OR_CARE_REFERENCE","text":"11223344556","start":234,"end":245,"sensitivity":"professional_reference","required":true,"acceptable_entity_types":["NL_BIG_NUMBER","NL_HEALTHCARE_REFERENCE"],"notes":"BIG-number-like value."},{"id":"L009","entity_class":"PERSON","text":"Hassan El Amrani","start":292,"end":308,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PERSON"],"notes":"Person name."},{"id":"L010","entity_class":"PERSON","text":"Mila van Dijk","start":312,"end":325,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PERSON"],"notes":"Person name."},{"id":"L011","entity_class":"EMAIL","text":"hassan.elamrani@example.test","start":346,"end":374,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["EMAIL_ADDRESS"],"notes":"Reserved synthetic email domain."},{"id":"L012","entity_class":"PHONE","text":"06 55667788","start":378,"end":389,"sensitivity":"direct_identifier","required":true,"acceptable_entity_types":["PHONE_NUMBER","NL_PHONE_NUMBER"],"notes":"Synthetic mobile number."}],"preserve_terms":[{"term":"Kamer","start":0,"end":5,"reason":"care location context term"},{"term":"afdeling","start":20,"end":28,"reason":"care location context term"},{"term":"cliënt","start":80,"end":86,"reason":"care role/context term"},{"term":"arts","start":259,"end":263,"reason":"care role/context term"},{"term":"verpleegkundige","start":414,"end":429,"reason":"care role/context term"},{"term":"zorgmedewerker","start":436,"end":450,"reason":"care role/context term"}],"known_traps":[{"text":"Kamer C-118","trap_type":"care_location_context_needs_review","notes":"Room references may be identifying in care context and need explicit handling."},{"text":"afdeling Rozenhof 3","trap_type":"care_location_context_needs_review","notes":"Department references may be identifying depending on context."},{"text":"cliënt","trap_type":"care_role_word_must_not_be_person","notes":"Generic care role word should remain readable."}]}