Update default.yaml
This commit is contained in:
163
default.yaml
163
default.yaml
@@ -16,166 +16,3 @@ ner_model_configuration:
|
|||||||
labels_to_ignore:
|
labels_to_ignore:
|
||||||
- MISC
|
- MISC
|
||||||
|
|
||||||
confidence_threshold:
|
|
||||||
default: 0.6
|
|
||||||
EMAIL_ADDRESS: 0.8
|
|
||||||
PHONE_NUMBER: 0.8
|
|
||||||
PERSON: 0.85
|
|
||||||
LOCATION: 0.75
|
|
||||||
MONEY: 0.85
|
|
||||||
|
|
||||||
# Configuration des détecteurs (Recognizers)
|
|
||||||
# Ici, nous pointons directement vers le module Python que nous venons de créer.
|
|
||||||
# Presidio chargera tous les recognizers définis dans ce module.
|
|
||||||
recognizer_registry:
|
|
||||||
- default # Charger les détecteurs par défaut de Presidio
|
|
||||||
- custom_recognizers # Référence au dossier/module Python 'custom_recognizers'
|
|
||||||
|
|
||||||
# Liste des mots à ignorer (faux positifs courants)
|
|
||||||
allow_list:
|
|
||||||
# Termes contractuels/juridiques/financiers génériques souvent mal labellisés
|
|
||||||
- text: Contrat
|
|
||||||
type: LOCATION
|
|
||||||
- text: contrat
|
|
||||||
type: LOCATION
|
|
||||||
- text: Contrats
|
|
||||||
type: LOCATION
|
|
||||||
- text: Document
|
|
||||||
type: LOCATION
|
|
||||||
- text: document
|
|
||||||
type: LOCATION
|
|
||||||
- text: Société
|
|
||||||
type: PERSON
|
|
||||||
- text: Investisseur
|
|
||||||
type: PERSON
|
|
||||||
- text: Montant
|
|
||||||
type: LOCATION
|
|
||||||
- text: Prêt
|
|
||||||
type: LOCATION
|
|
||||||
- text: Intérêt
|
|
||||||
type: LOCATION
|
|
||||||
- text: Intérêts
|
|
||||||
type: LOCATION
|
|
||||||
- text: Partie
|
|
||||||
type: LOCATION
|
|
||||||
- text: Parties
|
|
||||||
type: PERSON
|
|
||||||
- text: Annexe
|
|
||||||
type: LOCATION
|
|
||||||
- text: Remboursement
|
|
||||||
type: LOCATION
|
|
||||||
- text: Conversion
|
|
||||||
type: LOCATION
|
|
||||||
- text: Financement
|
|
||||||
type: LOCATION
|
|
||||||
- text: Sortie
|
|
||||||
type: LOCATION
|
|
||||||
- text: "Juste Valeur Marchande"
|
|
||||||
type: PERSON
|
|
||||||
- text: Échéance
|
|
||||||
type: LOCATION
|
|
||||||
- text: Clause
|
|
||||||
type: LOCATION
|
|
||||||
- text: Clauses
|
|
||||||
type: LOCATION
|
|
||||||
- text: Principe
|
|
||||||
type: LOCATION
|
|
||||||
- text: Coûts
|
|
||||||
type: PERSON
|
|
||||||
- text: Notifications
|
|
||||||
type: LOCATION
|
|
||||||
- text: Article
|
|
||||||
type: LOCATION
|
|
||||||
- text: Paragraphe
|
|
||||||
type: LOCATION
|
|
||||||
- text: Directeur
|
|
||||||
type: LOCATION
|
|
||||||
- text: Gérant
|
|
||||||
type: LOCATION
|
|
||||||
- text: Président
|
|
||||||
type: LOCATION
|
|
||||||
- text: DocuSign
|
|
||||||
type: PERSON
|
|
||||||
- text: SPRL
|
|
||||||
type: ORG
|
|
||||||
|
|
||||||
# Termes financiers courants
|
|
||||||
- text: Euro
|
|
||||||
type: LOCATION
|
|
||||||
- text: EUR
|
|
||||||
type: LOCATION
|
|
||||||
- text: Euros
|
|
||||||
type: LOCATION
|
|
||||||
- text: Pourcentage
|
|
||||||
type: LOCATION
|
|
||||||
- text: Taux
|
|
||||||
type: LOCATION
|
|
||||||
- text: Valeur
|
|
||||||
type: LOCATION
|
|
||||||
- text: Prix
|
|
||||||
type: LOCATION
|
|
||||||
|
|
||||||
# Termes juridiques
|
|
||||||
- text: Loi
|
|
||||||
type: LOCATION
|
|
||||||
- text: Code
|
|
||||||
type: LOCATION
|
|
||||||
- text: Règlement
|
|
||||||
type: LOCATION
|
|
||||||
- text: Décret
|
|
||||||
type: LOCATION
|
|
||||||
- text: Arrêté
|
|
||||||
type: LOCATION
|
|
||||||
|
|
||||||
# Mois et dates génériques
|
|
||||||
- text: Janvier
|
|
||||||
type: LOCATION
|
|
||||||
- text: Février
|
|
||||||
type: LOCATION
|
|
||||||
- text: Mars
|
|
||||||
type: LOCATION
|
|
||||||
- text: Avril
|
|
||||||
type: LOCATION
|
|
||||||
- text: Mai
|
|
||||||
type: LOCATION
|
|
||||||
- text: Juin
|
|
||||||
type: LOCATION
|
|
||||||
- text: Juillet
|
|
||||||
type: LOCATION
|
|
||||||
- text: Août
|
|
||||||
type: LOCATION
|
|
||||||
- text: Septembre
|
|
||||||
type: LOCATION
|
|
||||||
- text: Octobre
|
|
||||||
type: LOCATION
|
|
||||||
- text: Novembre
|
|
||||||
type: LOCATION
|
|
||||||
- text: Décembre
|
|
||||||
|
|
||||||
# Configuration des transformations d'anonymisation
|
|
||||||
anonymizer_config:
|
|
||||||
default_anonymizers:
|
|
||||||
PERSON: replace
|
|
||||||
EMAIL_ADDRESS: replace
|
|
||||||
PHONE_NUMBER: replace
|
|
||||||
BE_NATIONAL_REGISTER_NUMBER: replace
|
|
||||||
BE_ENTERPRISE_NUMBER: replace
|
|
||||||
FR_SOCIAL_SECURITY_NUMBER: replace
|
|
||||||
IBAN: replace
|
|
||||||
BE_BANK_ACCOUNT: replace
|
|
||||||
MONEY: replace
|
|
||||||
LOCATION: replace
|
|
||||||
ORG: replace
|
|
||||||
|
|
||||||
replacements:
|
|
||||||
PERSON: "<PERSON>"
|
|
||||||
EMAIL_ADDRESS: "<EMAIL_ADDRESS>"
|
|
||||||
PHONE_NUMBER: "<PHONE_NUMBER>"
|
|
||||||
BE_NATIONAL_REGISTER_NUMBER: "<BE_NATIONAL_REGISTER_NUMBER>"
|
|
||||||
BE_ENTERPRISE_NUMBER: "<BE_ENTERPRISE_NUMBER>"
|
|
||||||
FR_SOCIAL_SECURITY_NUMBER: "<FR_SOCIAL_SECURITY_NUMBER>"
|
|
||||||
IBAN: "<IBAN>"
|
|
||||||
BE_BANK_ACCOUNT: "<COMPTE_BANCAIRE_BE>"
|
|
||||||
MONEY: "<MONEY>"
|
|
||||||
LOCATION: "<LOCATION>"
|
|
||||||
ORG: "<ORGANIZATION>"
|
|
||||||
|
|||||||
Reference in New Issue
Block a user