573 lines
12 KiB
JSON
573 lines
12 KiB
JSON
{
|
|
"source_file": "CRH 23132490.pdf",
|
|
"total_replacements": 479,
|
|
"regex_replacements": 237,
|
|
"ner_replacements": 16,
|
|
"sweep_replacements": 226,
|
|
"entities_found": [
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10110601324",
|
|
"replacement": "[RPPS_1]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101096005",
|
|
"replacement": "[RPPS_2]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100163277",
|
|
"replacement": "[RPPS_3]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101639580",
|
|
"replacement": "[RPPS_4]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10004606595",
|
|
"replacement": "[RPPS_5]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101480506",
|
|
"replacement": "[RPPS_6]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100858090",
|
|
"replacement": "[RPPS_7]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102268702",
|
|
"replacement": "[RPPS_8]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10102272209",
|
|
"replacement": "[RPPS_9]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10002815024",
|
|
"replacement": "[RPPS_11]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10100817005",
|
|
"replacement": "[RPPS_12]",
|
|
"source": "regex",
|
|
"category": "rpps"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "code_barre"
|
|
},
|
|
{
|
|
"original": "10101856135",
|
|
"replacement": "[RPPS_10]",
|
|
"source": "regex",
|
|
"category": "code_barre"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "secr.medint@ch-cotebasque.fr",
|
|
"replacement": "[EMAIL_1]",
|
|
"source": "regex",
|
|
"category": "email"
|
|
},
|
|
{
|
|
"original": "Kappa",
|
|
"replacement": "[PERSONNE_1]",
|
|
"source": "ner",
|
|
"score": 0.840074360370636
|
|
},
|
|
{
|
|
"original": "[MEDECIN_47",
|
|
"replacement": "[PERSONNE_2]",
|
|
"source": "ner",
|
|
"score": 0.9252339005470276
|
|
},
|
|
{
|
|
"original": "[MEDECIN_50",
|
|
"replacement": "[PERSONNE_3]",
|
|
"source": "ner",
|
|
"score": 0.980678379535675
|
|
},
|
|
{
|
|
"original": "J. LOEB",
|
|
"replacement": "[PERSONNE_4]",
|
|
"source": "ner",
|
|
"score": 0.8677589893341064
|
|
},
|
|
{
|
|
"original": "[MEDECIN",
|
|
"replacement": "[PERSONNE_5]",
|
|
"source": "ner",
|
|
"score": 0.9780985116958618
|
|
},
|
|
{
|
|
"original": "Léa LOPEZ",
|
|
"replacement": "[MEDECIN_4]",
|
|
"source": "ner",
|
|
"score": 0.9967891573905945
|
|
},
|
|
{
|
|
"original": "[MEDEC",
|
|
"replacement": "[PERSONNE_6]",
|
|
"source": "ner",
|
|
"score": 0.9109337329864502
|
|
},
|
|
{
|
|
"original": "[MEDECIN_",
|
|
"replacement": "[PERSONNE_7]",
|
|
"source": "ner",
|
|
"score": 0.979133665561676
|
|
},
|
|
{
|
|
"original": "[MEDECIN_37",
|
|
"replacement": "[PERSONNE_8]",
|
|
"source": "ner",
|
|
"score": 0.9210373759269714
|
|
},
|
|
{
|
|
"original": "[MEDECIN_35",
|
|
"replacement": "[PERSONNE_9]",
|
|
"source": "ner",
|
|
"score": 0.9335954785346985
|
|
},
|
|
{
|
|
"original": "[MEDECIN_22",
|
|
"replacement": "[PERSONNE_10]",
|
|
"source": "ner",
|
|
"score": 0.9579775929450989
|
|
},
|
|
{
|
|
"original": "[MEDECIN_23",
|
|
"replacement": "[PERSONNE_11]",
|
|
"source": "ner",
|
|
"score": 0.9376325011253357
|
|
},
|
|
{
|
|
"original": "[MEDECIN_24",
|
|
"replacement": "[PERSONNE_12]",
|
|
"source": "ner",
|
|
"score": 0.9113510847091675
|
|
},
|
|
{
|
|
"original": "[MEDECIN_46",
|
|
"replacement": "[PERSONNE_13]",
|
|
"source": "ner",
|
|
"score": 0.968749463558197
|
|
},
|
|
{
|
|
"original": "[MEDECIN_49",
|
|
"replacement": "[PERSONNE_14]",
|
|
"source": "ner",
|
|
"score": 0.9341500401496887
|
|
},
|
|
{
|
|
"original": "LAROYE Parfait",
|
|
"replacement": "[PERSONNE_15]",
|
|
"source": "ner",
|
|
"score": 0.9883614778518677
|
|
}
|
|
]
|
|
} |