Files
anonymisation/tests/ground_truth/annotations/dataset_statistics.json

23 lines
399 B
JSON

{
"total_documents": 25,
"total_pages": 133,
"total_pii": 1167,
"by_type": {
"ETABLISSEMENT": 83,
"TEL": 193,
"NOM": 507,
"IPP": 25,
"ADRESSE": 79,
"CODE_POSTAL": 50,
"DATE_NAISSANCE": 114,
"EMAIL": 62,
"RPPS": 21,
"EPISODE": 18,
"VILLE": 5,
"AGE": 5,
"NIR": 2,
"DOSSIER": 3
},
"avg_pii_per_doc": 46.7,
"avg_pages_per_doc": 5.3
}