chore: add .gitignore

This commit is contained in:
dom
2026-03-05 00:37:34 +01:00
parent da34bdc8d7
commit d2e0fec97d
2087 changed files with 1485338 additions and 14 deletions

View File

@@ -0,0 +1,399 @@
{
"source_file": "CRH 23171519.pdf",
"total_replacements": 210,
"regex_replacements": 149,
"ner_replacements": 19,
"sweep_replacements": 42,
"entities_found": [
{
"original": "10100532760",
"replacement": "[RPPS_1]",
"source": "regex",
"category": "rpps"
},
{
"original": "10100532760",
"replacement": "[RPPS_1]",
"source": "regex",
"category": "rpps"
},
{
"original": "10100532760",
"replacement": "[RPPS_1]",
"source": "regex",
"category": "rpps"
},
{
"original": "10100532760",
"replacement": "[RPPS_1]",
"source": "regex",
"category": "rpps"
},
{
"original": "10002111572",
"replacement": "[CODE_BARRE_1]",
"source": "regex",
"category": "code_barre"
},
{
"original": "10002111572",
"replacement": "[CODE_BARRE_1]",
"source": "regex",
"category": "code_barre"
},
{
"original": "fprevost@ch-cotebasque.fr",
"replacement": "[EMAIL_1]",
"source": "regex",
"category": "email"
},
{
"original": "boui@ch-cotebasque.fr",
"replacement": "[EMAIL_2]",
"source": "regex",
"category": "email"
},
{
"original": "dnivet@ch-cotebasque.fr",
"replacement": "[EMAIL_3]",
"source": "regex",
"category": "email"
},
{
"original": "tkhuong-huu@ch-cotebasque.fr",
"replacement": "[EMAIL_4]",
"source": "regex",
"category": "email"
},
{
"original": "aguilngar@ch-cotebasque.fr",
"replacement": "[EMAIL_5]",
"source": "regex",
"category": "email"
},
{
"original": "fgoutorbe@ch-cotebasque.fr",
"replacement": "[EMAIL_6]",
"source": "regex",
"category": "email"
},
{
"original": "mcboudier@ch-cotebasque.fr",
"replacement": "[EMAIL_7]",
"source": "regex",
"category": "email"
},
{
"original": "mbrugel@ch-cotebasque.fr",
"replacement": "[EMAIL_8]",
"source": "regex",
"category": "email"
},
{
"original": "mboube@ch-cotebasque.fr",
"replacement": "[EMAIL_9]",
"source": "regex",
"category": "email"
},
{
"original": "faudemar@ch-cotebasque.fr",
"replacement": "[EMAIL_10]",
"source": "regex",
"category": "email"
},
{
"original": "fprevost@ch-cotebasque.fr",
"replacement": "[EMAIL_1]",
"source": "regex",
"category": "email"
},
{
"original": "boui@ch-cotebasque.fr",
"replacement": "[EMAIL_2]",
"source": "regex",
"category": "email"
},
{
"original": "dnivet@ch-cotebasque.fr",
"replacement": "[EMAIL_3]",
"source": "regex",
"category": "email"
},
{
"original": "tkhuong-huu@ch-cotebasque.fr",
"replacement": "[EMAIL_4]",
"source": "regex",
"category": "email"
},
{
"original": "aguilngar@ch-cotebasque.fr",
"replacement": "[EMAIL_5]",
"source": "regex",
"category": "email"
},
{
"original": "fgoutorbe@ch-cotebasque.fr",
"replacement": "[EMAIL_6]",
"source": "regex",
"category": "email"
},
{
"original": "mcboudier@ch-cotebasque.fr",
"replacement": "[EMAIL_7]",
"source": "regex",
"category": "email"
},
{
"original": "mbrugel@ch-cotebasque.fr",
"replacement": "[EMAIL_8]",
"source": "regex",
"category": "email"
},
{
"original": "mboube@ch-cotebasque.fr",
"replacement": "[EMAIL_9]",
"source": "regex",
"category": "email"
},
{
"original": "faudemar@ch-cotebasque.fr",
"replacement": "[EMAIL_10]",
"source": "regex",
"category": "email"
},
{
"original": "fprevost@ch-cotebasque.fr",
"replacement": "[EMAIL_1]",
"source": "regex",
"category": "email"
},
{
"original": "boui@ch-cotebasque.fr",
"replacement": "[EMAIL_2]",
"source": "regex",
"category": "email"
},
{
"original": "dnivet@ch-cotebasque.fr",
"replacement": "[EMAIL_3]",
"source": "regex",
"category": "email"
},
{
"original": "tkhuong-huu@ch-cotebasque.fr",
"replacement": "[EMAIL_4]",
"source": "regex",
"category": "email"
},
{
"original": "aguilngar@ch-cotebasque.fr",
"replacement": "[EMAIL_5]",
"source": "regex",
"category": "email"
},
{
"original": "fgoutorbe@ch-cotebasque.fr",
"replacement": "[EMAIL_6]",
"source": "regex",
"category": "email"
},
{
"original": "mcboudier@ch-cotebasque.fr",
"replacement": "[EMAIL_7]",
"source": "regex",
"category": "email"
},
{
"original": "mbrugel@ch-cotebasque.fr",
"replacement": "[EMAIL_8]",
"source": "regex",
"category": "email"
},
{
"original": "mboube@ch-cotebasque.fr",
"replacement": "[EMAIL_9]",
"source": "regex",
"category": "email"
},
{
"original": "faudemar@ch-cotebasque.fr",
"replacement": "[EMAIL_10]",
"source": "regex",
"category": "email"
},
{
"original": "fprevost@ch-cotebasque.fr",
"replacement": "[EMAIL_1]",
"source": "regex",
"category": "email"
},
{
"original": "boui@ch-cotebasque.fr",
"replacement": "[EMAIL_2]",
"source": "regex",
"category": "email"
},
{
"original": "dnivet@ch-cotebasque.fr",
"replacement": "[EMAIL_3]",
"source": "regex",
"category": "email"
},
{
"original": "tkhuong-huu@ch-cotebasque.fr",
"replacement": "[EMAIL_4]",
"source": "regex",
"category": "email"
},
{
"original": "aguilngar@ch-cotebasque.fr",
"replacement": "[EMAIL_5]",
"source": "regex",
"category": "email"
},
{
"original": "fgoutorbe@ch-cotebasque.fr",
"replacement": "[EMAIL_6]",
"source": "regex",
"category": "email"
},
{
"original": "mcboudier@ch-cotebasque.fr",
"replacement": "[EMAIL_7]",
"source": "regex",
"category": "email"
},
{
"original": "mbrugel@ch-cotebasque.fr",
"replacement": "[EMAIL_8]",
"source": "regex",
"category": "email"
},
{
"original": "mboube@ch-cotebasque.fr",
"replacement": "[EMAIL_9]",
"source": "regex",
"category": "email"
},
{
"original": "faudemar@ch-cotebasque.fr",
"replacement": "[EMAIL_10]",
"source": "regex",
"category": "email"
},
{
"original": "A.GUILNGAR",
"replacement": "[MEDECIN]",
"source": "ner",
"score": 0.9128968119621277
},
{
"original": "C. CAZELLES-BOUDIER",
"replacement": "[MEDECIN_37]",
"source": "ner",
"score": 0.9121089577674866
},
{
"original": "M. [PATIENT",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.9687307476997375
},
{
"original": "[MEDECIN_29",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9634112119674683
},
{
"original": "J. LOEB",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.8792789578437805
},
{
"original": "[MEDECIN",
"replacement": "[PERSONNE_4]",
"source": "ner",
"score": 0.9849982261657715
},
{
"original": "[MEDECIN_37",
"replacement": "[PERSONNE_5]",
"source": "ner",
"score": 0.940574049949646
},
{
"original": "Thomas Grellety",
"replacement": "[MEDECIN_1]",
"source": "ner",
"score": 0.9969271421432495
},
{
"original": "A.GUILNGAR A",
"replacement": "[PERSONNE_6]",
"source": "ner",
"score": 0.8633033633232117
},
{
"original": "PATIENT",
"replacement": "[PERSONNE_7]",
"source": "ner",
"score": 0.8335493206977844
},
{
"original": "[MEDECIN_",
"replacement": "[PERSONNE_8]",
"source": "ner",
"score": 0.9617339968681335
},
{
"original": "[MEDECIN_33",
"replacement": "[PERSONNE_9]",
"source": "ner",
"score": 0.9225618243217468
},
{
"original": "[MEDECIN_36",
"replacement": "[PERSONNE_10]",
"source": "ner",
"score": 0.9599934220314026
},
{
"original": "M. [MEDECIN",
"replacement": "[PERSONNE_11]",
"source": "ner",
"score": 0.9767007827758789
},
{
"original": "LABENNE Suzy",
"replacement": "[PATIENT_1]",
"source": "ner",
"score": 0.9807483553886414
},
{
"original": "ATXIKI",
"replacement": "[ADRESSE]",
"source": "ner",
"score": 0.8165812492370605
},
{
"original": "[MEDECIN_1",
"replacement": "[PERSONNE_12]",
"source": "ner",
"score": 0.9403961896896362
},
{
"original": "[PATIENT_1",
"replacement": "[PERSONNE_13]",
"source": "ner",
"score": 0.8430862426757812
},
{
"original": "[PATIENT_5",
"replacement": "[PERSONNE_14]",
"source": "ner",
"score": 0.9390151500701904
}
]
}

View File

@@ -0,0 +1,33 @@
{
"source_file": "CR RADIO INTERVENTION 23171519.pdf",
"total_replacements": 19,
"regex_replacements": 9,
"ner_replacements": 2,
"sweep_replacements": 8,
"entities_found": [
{
"original": "286074000102496",
"replacement": "[NIR_1]",
"source": "regex",
"category": "nir"
},
{
"original": "40001",
"replacement": "[LIEU_NAISS_1]",
"source": "regex",
"category": "lieu_naissance"
},
{
"original": "[MEDECIN_",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.9388558864593506
},
{
"original": "SOULAT Marie",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.976779580116272
}
]
}

View File

@@ -0,0 +1,309 @@
{
"source_file": "trackare-20009129-23171519_20009129_23171519.pdf",
"total_replacements": 360,
"regex_replacements": 104,
"ner_replacements": 20,
"sweep_replacements": 236,
"entities_found": [
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "20009129",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23171519",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "640000162",
"replacement": "[FINESS]",
"source": "regex",
"category": "finess"
},
{
"original": "AIRE SUR L ADOUR",
"replacement": "[LIEU_NAISS_1]",
"source": "regex",
"category": "lieu_naissance"
},
{
"original": "Yohan BENARD",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.9982704520225525
},
{
"original": "Marion BEVIERE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.994347095489502
},
{
"original": "KUTZ Rainer",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.9965077638626099
},
{
"original": "Maylis Réalisé",
"replacement": "[PERSONNE_4]",
"source": "ner",
"score": 0.9965002536773682
},
{
"original": "Caroline Signé",
"replacement": "[PERSONNE_5]",
"source": "ner",
"score": 0.9967696070671082
},
{
"original": "OXYNORMORO",
"replacement": "[PERSONNE_6]",
"source": "ner",
"score": 0.9950122833251953
},
{
"original": "Caroline",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9556587338447571
},
{
"original": "MARGUERITE S",
"replacement": "[PERSONNE_7]",
"source": "ner",
"score": 0.9737850427627563
},
{
"original": "KUTZ Rainer Réalisé",
"replacement": "[PERSONNE_8]",
"source": "ner",
"score": 0.9936642050743103
},
{
"original": "Maylis",
"replacement": "[PERSONNE]",
"source": "ner",
"score": 0.9764008522033691
},
{
"original": "GELULE",
"replacement": "[MEDECIN]",
"source": "ner",
"score": 0.8406248092651367
},
{
"original": "DISP Caroline",
"replacement": "[PERSONNE_9]",
"source": "ner",
"score": 0.8761375546455383
},
{
"original": "MARGUERITE",
"replacement": "[MEDECIN]",
"source": "ner",
"score": 0.8951443433761597
},
{
"original": "Mathilde S",
"replacement": "[PERSONNE_10]",
"source": "ner",
"score": 0.8007203340530396
},
{
"original": "MEDECIN_",
"replacement": "[PERSONNE_11]",
"source": "ner",
"score": 0.9302533864974976
},
{
"original": "[MEDECIN_24",
"replacement": "[PERSONNE_12]",
"source": "ner",
"score": 0.9388594627380371
},
{
"original": "H Cailliez",
"replacement": "[PERSONNE_13]",
"source": "ner",
"score": 0.9918519854545593
},
{
"original": "[MEDECIN_",
"replacement": "[PERSONNE_14]",
"source": "ner",
"score": 0.9613234400749207
},
{
"original": "[MEDECIN",
"replacement": "[PERSONNE_15]",
"source": "ner",
"score": 0.9305795431137085
},
{
"original": "Jean Monnet",
"replacement": "[PERSONNE_16]",
"source": "ner",
"score": 0.9847896695137024
}
]
}