Dom/anonymisation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: Désactivation NOM_EXTRACTED et *_GLOBAL - Précision 18.97% → 88.27% (+69.3pts)

Browse Source
This commit is contained in:
Domi31tls
2026-03-02 11:15:43 +01:00
parent dfa45041d7
commit 70ff0b9e12
89 changed files with 1261 additions and 6182 deletions
Download Patch File Download Diff File Expand all files Collapse all files

214
tests/ground_truth/quality_evaluation/baseline_quality_evaluation.json
View File

@@ -2,11 +2,11 @@
"evaluation_date": "2026-03-02",
"total_documents": 25,
"global_metrics": {
"precision": 0.1897,
"precision": 0.8827,
"recall": 1.0,
"f1_score": 0.3189,
"f1_score": 0.9377,
"true_positives": 1159,
"false_positives": 4951,
"false_positives": 154,
"false_negatives": 0
},
"by_type": {
@@ -42,38 +42,6 @@
"false_positives": 0,
"false_negatives": 0
},
"NOM_EXTRACTED": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 3846,
"false_negatives": 0
},
"NOM_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 670,
"false_negatives": 0
},
"ETAB_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 36,
"false_negatives": 0
},
"TEL_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 77,
"false_negatives": 0
},
"ADRESSE": {
"precision": 0.878,
"recall": 1.0,
@@ -98,30 +66,6 @@
"false_positives": 0,
"false_negatives": 0
},
"ADRESSE_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 55,
"false_negatives": 0
},
"CODE_POSTAL_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 39,
"false_negatives": 0
},
"DATE_NAISSANCE_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 20,
"false_negatives": 0
},
"EMAIL": {
"precision": 1.0,
"recall": 1.0,
@@ -146,30 +90,6 @@
"false_positives": 106,
"false_negatives": 0
},
"EMAIL_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 28,
"false_negatives": 0
},
"RPPS_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 7,
"false_negatives": 0
},
"EPISODE_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 9,
"false_negatives": 0
},
"VILLE": {
"precision": 0.2,
"recall": 1.0,
@@ -178,14 +98,6 @@
"false_positives": 20,
"false_negatives": 0
},
"VILLE_GLOBAL": {
"precision": 0.0,
"recall": 0.0,
"f1_score": 0.0,
"true_positives": 0,
"false_positives": 10,
"false_negatives": 0
},
"AGE": {
"precision": 1.0,
"recall": 1.0,
@@ -214,29 +126,29 @@
"per_document": [
{
"pdf": "001_simple_unknown_BACTERIO_23018396",
"precision": 0.2326,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.3774,
"f1_score": 1.0,
"true_positives": 10,
"false_positives": 33,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "002_simple_unknown_bacterio_476_23159413",
"precision": 0.234,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.3793,
"f1_score": 1.0,
"true_positives": 11,
"false_positives": 36,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "003_simple_compte_rendu_CRO_23155084",
"precision": 0.16,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.2759,
"f1_score": 1.0,
"true_positives": 4,
"false_positives": 21,
"false_positives": 0,
"false_negatives": 0
},
{
@@ -250,65 +162,65 @@
},
{
"pdf": "005_simple_compte_rendu_CRH_23155836",
"precision": 0.4429,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.6139,
"f1_score": 1.0,
"true_positives": 62,
"false_positives": 78,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "008_simple_trackare_trackare-14004105-23202435_14004105_23202435",
"precision": 0.1899,
"precision": 0.5769,
"recall": 1.0,
"f1_score": 0.3191,
"f1_score": 0.7317,
"true_positives": 15,
"false_positives": 64,
"false_positives": 11,
"false_negatives": 0
},
{
"pdf": "009_simple_compte_rendu_CRO_23051225",
"precision": 0.25,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.4,
"f1_score": 1.0,
"true_positives": 8,
"false_positives": 24,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "010_simple_anapath_ANAPATH_23217289",
"precision": 0.2549,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.4062,
"f1_score": 1.0,
"true_positives": 13,
"false_positives": 38,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "011_moyen_compte_rendu_CRH_23080179",
"precision": 0.3158,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.48,
"f1_score": 1.0,
"true_positives": 12,
"false_positives": 26,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "012_moyen_compte_rendu_CRH_692_23200418",
"precision": 0.297,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.458,
"f1_score": 1.0,
"true_positives": 30,
"false_positives": 71,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "013_moyen_compte_rendu_363_23085243_CRO",
"precision": 0.2025,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.3368,
"f1_score": 1.0,
"true_positives": 32,
"false_positives": 126,
"false_positives": 0,
"false_negatives": 0
},
{
@@ -322,20 +234,20 @@
},
{
"pdf": "015_moyen_unknown_CONSULTATION_ANESTHESISTE_23139653",
"precision": 0.28,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.4375,
"f1_score": 1.0,
"true_positives": 7,
"false_positives": 18,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "016_moyen_compte_rendu_CRH_23149905",
"precision": 0.477,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.6459,
"f1_score": 1.0,
"true_positives": 114,
"false_positives": 125,
"false_positives": 0,
"false_negatives": 0
},
{
@@ -349,20 +261,20 @@
},
{
"pdf": "018_moyen_compte_rendu_CRH_23042753",
"precision": 0.5279,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.691,
"f1_score": 1.0,
"true_positives": 123,
"false_positives": 110,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "019_moyen_compte_rendu_CRO_332_23049003",
"precision": 0.3793,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.55,
"f1_score": 1.0,
"true_positives": 55,
"false_positives": 90,
"false_positives": 0,
"false_negatives": 0
},
{
@@ -385,56 +297,56 @@
},
{
"pdf": "022_moyen_compte_rendu_cro2_516_23187028",
"precision": 0.1379,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.2424,
"f1_score": 1.0,
"true_positives": 4,
"false_positives": 25,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "023_complexe_compte_rendu_CRH_23102610",
"precision": 0.6203,
"precision": 1.0,
"recall": 1.0,
"f1_score": 0.7657,
"f1_score": 1.0,
"true_positives": 379,
"false_positives": 232,
"false_positives": 0,
"false_negatives": 0
},
{
"pdf": "024_complexe_trackare_trackare-17001141-23066188_17001141_23066188",
"precision": 0.0689,
"precision": 0.6463,
"recall": 1.0,
"f1_score": 0.129,
"f1_score": 0.7852,
"true_positives": 53,
"false_positives": 716,
"false_positives": 29,
"false_negatives": 0
},
{
"pdf": "025_complexe_trackare_trackare-02016820-23095226_02016820_23095226",
"precision": 0.0643,
"precision": 0.6857,
"recall": 1.0,
"f1_score": 0.1209,
"f1_score": 0.8136,
"true_positives": 96,
"false_positives": 1396,
"false_positives": 44,
"false_negatives": 0
},
{
"pdf": "026_complexe_trackare_trackare-15000536-23074384_15000536_23074384",
"precision": 0.0766,
"precision": 0.6695,
"recall": 1.0,
"f1_score": 0.1422,
"f1_score": 0.802,
"true_positives": 79,
"false_positives": 953,
"false_positives": 39,
"false_negatives": 0
},
{
"pdf": "027_complexe_trackare_trackare-10027557-23183041_10027557_23183041",
"precision": 0.0633,
"precision": 0.6265,
"recall": 1.0,
"f1_score": 0.1191,
"f1_score": 0.7704,
"true_positives": 52,
"false_positives": 769,
"false_positives": 31,
"false_negatives": 0
}
]