diff --git a/Dockerfile.analyzer b/Dockerfile.analyzer
index ebb1d47..7371432 100644
--- a/Dockerfile.analyzer
+++ b/Dockerfile.analyzer
@@ -31,7 +31,7 @@ COPY . /app/
 # Définir la variable d'environnement pour que Presidio trouve notre fichier de configuration
 
 # Dit à Presidio : "Ton fichier de config est ici"
-ENV PRESIDIO_ANALYZER_CONFIG_FILE=/app/conf/default.yaml
+ENV PRESIDIO_ANALYZER_CONFIG_FILE=/app/conf/main.yaml
 
 # Exposer le port que Gunicorn va utiliser
 EXPOSE 5001
diff --git a/app.py b/app.py
index bbcbac8..4ea6fd6 100644
--- a/app.py
+++ b/app.py
@@ -1,59 +1,83 @@
 import os
-import re
 import logging
+import re
+import yaml
 from flask import Flask, request, jsonify, make_response
-
 from presidio_analyzer import AnalyzerEngineProvider
+from config_loader import ConfigLoader
+from presidio_anonymizer import AnonymizerEngine
+from presidio_anonymizer.entities import OperatorConfig
+from entity_refiners import EntityRefinerManager
+from pipeline_manager import AnalysisPipeline
 
+# Initialisation logger
 logging.basicConfig(level=logging.INFO,
                     format="%(asctime)s - %(name)s - %(levelname)s - %(message)s")
 logger = logging.getLogger(__name__)
 
 app = Flask(__name__)
 
-# Chargement du moteur 
-
+refiner_manager = EntityRefinerManager()
 analyzer = None
+allow_list_terms = set()
+
 try:
-    logger.info("--- Presidio Analyzer Service Starting ---")
-    CONFIG_FILE_PATH = os.environ.get("PRESIDIO_ANALYZER_CONFIG_FILE", "conf/default.yaml")
-    provider = AnalyzerEngineProvider(analyzer_engine_conf_file=CONFIG_FILE_PATH)
-    analyzer = provider.create_engine()
+    logger.info("--- Presidio Analyzer Service Starting (Architecture Modulaire) ---")
+    config_loader = ConfigLoader()
+    try:
+        config = config_loader.load_config("main.yaml")
+        logger.info("✅ Configuration modulaire chargée avec succès")
+
+        allow_list_terms = set(term.lower().strip() for term in config.get('allow_list', []))
+        logger.info(f"✅ Allow list chargée avec {len(allow_list_terms)} termes")
+
+        recognizers_count = len(config.get('recognizer_registry', {}).get('recognizers', []))
+        logger.info(f"📊 Nombre de recognizers chargés: {recognizers_count}")
+
+        import tempfile
+
+        # Écriture fichier temporaire config pour Presidio
+        with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False, encoding='utf-8') as tmp_file:
+            yaml.dump(config, tmp_file, default_flow_style=False, allow_unicode=True)
+            temp_config_path = tmp_file.name
+
+        with open(temp_config_path, 'r', encoding='utf-8') as f:
+            temp_content = f.read()
+            logger.info(f"🔍 Contenu du fichier temporaire COMPLET:\n{temp_content[:1000]}")
+
+        if 'nlp_configuration' in config:
+            logger.info("✅ nlp_configuration trouvée")
+        else:
+            logger.warning("❌ nlp_configuration MANQUANTE dans la config finale")
+
+        provider = AnalyzerEngineProvider(analyzer_engine_conf_file=temp_config_path)
+        analyzer = provider.create_engine()
+        os.unlink(temp_config_path)
+
+    except Exception as e:
+        logger.error(f"❌ Erreur avec la config modulaire: {e}")
+        logger.warning("🔄 Fallback vers default.yaml")
+        CONFIG_FILE_PATH = os.environ.get("PRESIDIO_ANALYZER_CONFIG_FILE", "conf/default.yaml")
+        provider = AnalyzerEngineProvider(analyzer_engine_conf_file=CONFIG_FILE_PATH)
+        analyzer = provider.create_engine()
+
     logger.info(f"Analyzer ready. Languages: {analyzer.supported_languages}")
+
 except Exception as e:
     logger.exception("Error during AnalyzerEngine initialization.")
     analyzer = None
 
 
-# Test Temporaire pour les Regex via du Python directement
-
-IBAN_REGEX = re.compile(r"\b[A-Z]{2}[0-9]{2}(?:\s[0-9]{4}){3}\b", re.IGNORECASE)
-
-
-IPV4_REGEX = re.compile(
-    r"\b(?:(?:25[0-5]|2[0-4][0-9]|1\d{2}|[1-9]?\d)\.){3}"
-    r"(?:25[0-5]|2[0-4][0-9]|1\d{2}|[1-9]?\d)\b"
-)
-
-# Liste Temporaire en surcouche des labels/phrases à exclure d’anonymisation 
-
-IGNORE_LABELS = {
-    "témoins",
-    "témoins clés",
-    "coordonnées",
-    "coordonnées bancaires",
-    "contexte financier",
-    "données sensibles",
-    "contexte",
-    "montrent",
-    "montrent des",
-    "montrent des irrégularités",
-    "bénéficiaire",
-}
-
 def normalize_label(text: str) -> str:
-    return text.strip().lower()
 
+    cleaned = re.sub(r'[^\w\s]', '', text.strip().lower())
+    return cleaned
+
+
+# Remplacer ligne 18
+pipeline = AnalysisPipeline()
+
+# Modifier la fonction analyze_text (lignes 73-105)
 @app.route("/analyze", methods=["POST"])
 def analyze_text():
     if not analyzer:
@@ -67,62 +91,182 @@ def analyze_text():
         if not text_to_analyze:
             return jsonify({"error": "text field is missing or empty"}), 400
 
-        results = analyzer.analyze(text=text_to_analyze, language=language)
-
-        filtered_results = []
-        for res in results:
-            ent_text = text_to_analyze[res.start:res.end].strip()
-            ent_text_norm = normalize_label(ent_text)
-
-            if ent_text_norm in IGNORE_LABELS:
-                logger.debug(f"Skipping anonymization of label: '{ent_text}'")
-                continue
-
-            # Recadrage IBAN 
-
-            if res.entity_type == "IBAN":
-                match = IBAN_REGEX.search(ent_text)
-                if match:
-                    true_iban = match.group(0)
-                    start_offset = ent_text.find(true_iban)
-                    if start_offset != -1:
-                        old_start, old_end = res.start, res.end
-                        res.start += start_offset
-                        res.end = res.start + len(true_iban)
-                        logger.debug(f"Adjusted IBAN span: {old_start}-{old_end} => {res.start}-{res.end}")
-                    else:
-                        logger.warning(f"IBAN regex match but cannot find substring position: '{ent_text}'")
-                else:
-                    logger.warning(f"Invalid IBAN detected, skipping: '{ent_text}'")
-                    continue
-
-            # Recadrage IP_ADDRESS 
-
-            if res.entity_type == "IP_ADDRESS":
-                match = IPV4_REGEX.search(ent_text)
-                if match:
-                    true_ip = match.group(0)
-                    start_offset = ent_text.find(true_ip)
-                    if start_offset != -1:
-                        old_start, old_end = res.start, res.end
-                        res.start += start_offset
-                        res.end = res.start + len(true_ip)
-                        logger.debug(f"Adjusted IP span: {old_start}-{old_end} => {res.start}-{res.end}")
-                    else:
-                        logger.warning(f"IP regex match but cannot find substring position: '{ent_text}'")
-                else:
-                    logger.warning(f"Invalid IP detected, skipping: '{ent_text}'")
-                    continue
-
-            filtered_results.append(res)
-
-        # Résultat nettoyé
-        response_data = [res.to_dict() for res in filtered_results]
+        # Analyse brute
+        raw_results = analyzer.analyze(text=text_to_analyze, language=language)
+        
+        # Pipeline modulaire complet
+        final_results = pipeline.process(text_to_analyze, raw_results, allow_list_terms)
+        
+        response_data = [res.to_dict() for res in final_results]
         return make_response(jsonify(response_data), 200)
 
     except Exception as e:
         logger.exception("Error processing analysis")
         return jsonify({"error": str(e)}), 500
 
+
+@app.route("/health", methods=["GET"])
+def health_check():
+    if analyzer:
+        return jsonify({
+            "status": "healthy",
+            "languages": analyzer.supported_languages,
+            "version": "2.0.0"
+        }), 200
+    else:
+        return jsonify({"status": "unhealthy", "error": "Analyzer not initialized"}), 503
+
+
+def load_replacements():
+    """Charge les configurations d'anonymisation depuis YAML"""
+    try:
+        config_path = "conf/anonymization/replacements.yaml"
+        if not os.path.exists(config_path):
+            logger.warning(f"❌ Fichier de configuration non trouvé: {config_path}")
+            return {}
+
+        with open(config_path, "r", encoding="utf-8") as f:
+            config = yaml.safe_load(f)
+
+        if not config:
+            logger.warning("❌ Fichier de configuration vide")
+            return {}
+
+        anonymizer_config = config.get("anonymizer_config", {})
+        replacements = anonymizer_config.get("replacements", {})
+
+        if not replacements:
+            logger.warning("❌ Aucun remplacement trouvé dans la configuration")
+            return {}
+
+        operators = {}
+        for entity_type, replacement_value in replacements.items():
+            try:
+                operators[entity_type] = OperatorConfig("replace", {"new_value": replacement_value})
+            except Exception as e:
+                logger.error(f"❌ Erreur lors création opérateur {entity_type}: {e}")
+                continue
+
+        logger.info(f"✅ Loaded {len(operators)} replacement operators from config")
+        return operators
+
+    except Exception as e:
+        logger.error(f"❌ Failed to load replacements config: {e}")
+        return {}
+
+
+# Initialisation anonymizer et opérateurs
+try:
+    anonymizer = AnonymizerEngine()
+    logger.info("✅ Anonymizer engine initialized successfully")
+    replacement_operators = load_replacements()
+    if replacement_operators:
+        logger.info(f"✅ Loaded {len(replacement_operators)} custom replacement operators")
+    else:
+        logger.warning("⚠️ Aucun opérateur remplacement chargé, fallback par défaut")
+        replacement_operators = {}
+
+except Exception as e:
+    logger.error(f"❌ Anonymizer initialization failed: {e}")
+    anonymizer = None
+    replacement_operators = {}
+
+
+@app.route("/anonymize", methods=["POST"])
+def anonymize_text():
+    logger.error("🚨 ENDPOINT /anonymize APPELÉ")
+
+    global anonymizer, replacement_operators
+
+    if anonymizer is None:
+        return jsonify({"error": "Anonymizer not initialized"}), 500
+
+    if not replacement_operators:
+        logger.warning("⚠️ replacement_operators non défini, rechargement...")
+        replacement_operators = load_replacements()
+
+    logger.info(f"🔍 Opérateurs disponibles: {list(replacement_operators.keys())}")
+
+    try:
+        data = request.get_json(force=True)
+        text_to_anonymize = data.get("text", "")
+        language = data.get("language", "fr")
+        mode = data.get("mode", "pii")
+
+        if not text_to_anonymize:
+            return jsonify({"error": "No text provided"}), 400
+
+        logger.info(f"🔍 Texte à anonymiser: '{text_to_anonymize}'")
+
+        entities_to_detect = get_entities_by_mode(mode) if 'get_entities_by_mode' in globals() else None
+
+        analyzer_results = analyzer.analyze(
+            text=text_to_anonymize,
+            language=language,
+            entities=entities_to_detect
+        )
+
+        logger.info(f"🔍 Entités détectées: {[(r.entity_type, text_to_anonymize[r.start:r.end], r.score) for r in analyzer_results]}")
+
+        filtered_results = []
+        for res in analyzer_results:
+            ent_text = text_to_anonymize[res.start:res.end].strip()
+            ent_text_norm = normalize_label(ent_text)
+
+            logger.info(f"🔍 Traitement entité: {res.entity_type} = '{ent_text}' (score: {res.score})")
+            logger.info(f"🔍 Allow list terms: {allow_list_terms}")
+
+            # Vérification améliorée de la allow list
+            ent_text_clean = re.sub(r'[^\w]', '', ent_text.strip().lower())
+            logger.info(f"🔍 Texte nettoyé: '{ent_text_clean}'")
+            
+            # Vérifier si le texte correspond exactement ou commence par un terme de la allow list
+            is_allowed = any(ent_text_clean == term or ent_text_clean.startswith(term) for term in allow_list_terms)
+            
+            if is_allowed:
+                logger.info(f"✅ Entité '{ent_text}' ignorée (dans allow list)")
+                continue
+
+            refined_positions = refiner_manager.refine_entity(text_to_anonymize, res.entity_type, res.start, res.end)
+            if refined_positions is None:
+                logger.info(f"❌ Entité {res.entity_type} supprimée par le refiner")
+                continue
+
+            res.start, res.end = refined_positions
+            filtered_results.append(res)
+            logger.info(f"✅ Entité {res.entity_type} conservée après refinement")
+
+        logger.info(f"🔍 Entités finales pour anonymisation: {[(r.entity_type, text_to_anonymize[r.start:r.end]) for r in filtered_results]}")
+
+        operators_to_use = replacement_operators if replacement_operators else {}
+        logger.info(f"🔍 Opérateurs utilisés: {list(operators_to_use.keys())}")
+
+        anonymized_result = anonymizer.anonymize(
+            text=text_to_anonymize,
+            analyzer_results=filtered_results,
+            operators=operators_to_use
+        )
+
+        logger.info(f"🔍 Résultat anonymisation: '{anonymized_result.text}'")
+
+        return jsonify({
+            "original_text": text_to_anonymize,
+            "anonymized_text": anonymized_result.text,
+            "entities_found": [
+                {
+                    "entity_type": result.entity_type,
+                    "start": result.start,
+                    "end": result.end,
+                    "score": result.score
+                } for result in filtered_results
+            ],
+            "mode": mode
+        })
+
+    except Exception as e:
+        logger.error(f"Error during anonymization: {e}")
+        return jsonify({"error": str(e)}), 500
+
+
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=5001)
diff --git a/conf/anonymization/allow_list.yaml b/conf/anonymization/allow_list.yaml
new file mode 100644
index 0000000..e6e4572
--- /dev/null
+++ b/conf/anonymization/allow_list.yaml
@@ -0,0 +1,19 @@
+# Liste blanche - termes à ne pas anonymiser
+allow_list:
+  # Références légales
+  - Loi
+  - Code
+  - Règlement
+  - Décret
+  - Arrêté
+  - BCE
+  - TVA
+  - IEC
+  - expert-comptable
+  # Termes financiers
+  - Euro
+  - EUR
+  - Euros
+  - Taux
+  - Valeur
+  - Prix
diff --git a/conf/anonymization/replacements.yaml b/conf/anonymization/replacements.yaml
new file mode 100644
index 0000000..1583740
--- /dev/null
+++ b/conf/anonymization/replacements.yaml
@@ -0,0 +1,82 @@
+# Configuration d'anonymisation complète
+anonymizer_config:
+  default_anonymizers:
+    # Entités génériques
+    PERSON: replace
+    LOCATION: replace
+    ORGANIZATION: replace
+    DATE_TIME: replace
+    MONEY: replace
+    EMAIL_ADDRESS: replace
+    IBAN: replace
+    IP_ADDRESS: replace
+
+    # PII Génériques - Données sensibles RGPD
+    HEALTH_DATA: replace
+    BIOMETRIC_DATA: replace
+    SEXUAL_ORIENTATION: replace
+    POLITICAL_OPINIONS: replace
+    RGPD_FINANCIAL_DATA: replace
+
+    # PII Belges
+    BE_ENTERPRISE_NUMBER: replace
+    BE_NATIONAL_REGISTER_NUMBER: replace
+    BE_PHONE_NUMBER: replace
+    BE_ADDRESS: replace
+    BE_ID_CARD: replace
+    BE_PASSPORT: replace
+
+    # PII Françaises
+    FR_SOCIAL_SECURITY_NUMBER: replace
+    FR_SIRET: replace
+    FR_ADDRESS: replace
+    FR_TAX_ID: replace
+    FR_BANK_ACCOUNT: replace
+    FR_ID_CARD: replace
+    FR_PASSPORT: replace
+    FR_DRIVER_LICENSE: replace
+
+    # Business
+    BE_PROFESSIONAL_ID: replace
+    MARKET_SHARE: replace
+
+  replacements:
+    # Entités génériques
+    PERSON: "[PERSONNE]"
+    LOCATION: "[LIEU]"
+    ORGANIZATION: "[ORGANISATION]"
+    DATE_TIME: "[DATE]"
+    MONEY: "[MONTANT]"
+    EMAIL_ADDRESS: "[EMAIL]"
+    IBAN: "[IBAN]"
+    IP_ADDRESS: "[ADRESSE_IP]"
+
+    # PII Génériques - Données sensibles RGPD
+    HEALTH_DATA: "[DONNEES_SANTE]"
+    BIOMETRIC_DATA: "[DONNEES_BIOMETRIQUES]"
+    SEXUAL_ORIENTATION: "[ORIENTATION_SEXUELLE]"
+    POLITICAL_OPINIONS: "[OPINIONS_POLITIQUES]"
+    RGPD_FINANCIAL_DATA: "[DONNEES_FINANCIERES]"
+
+    # PII Belges
+    BE_ENTERPRISE_NUMBER: "[ENTREPRISE_BELGE]"
+    BE_NATIONAL_REGISTER_NUMBER: "[NRN_BELGE]"
+    BE_PHONE_NUMBER: "[TELEPHONE_BE]"
+    BE_ADDRESS: "[ADRESSE_BELGE]"
+    BE_ID_CARD: "[CARTE_ID_BE]"
+    BE_PASSPORT: "[PASSEPORT_BE]"
+
+    # PII Françaises
+    FR_SOCIAL_SECURITY_NUMBER: "[NUM_SECU_FR]"
+    FR_SIRET: "[SIRET_FR]"
+    FR_ADDRESS: "[ADRESSE_FR]"
+    FR_TAX_ID: "[NUM_FISCAL_FR]"
+    FR_BANK_ACCOUNT: "[COMPTE_BANCAIRE_FR]"
+    FR_ID_CARD: "[CARTE_ID_FR]"
+    FR_PASSPORT: "[PASSEPORT_FR]"
+    FR_DRIVER_LICENSE: "[PERMIS_FR]"
+
+    # Business
+
+    BE_PROFESSIONAL_ID: "[ID_PROFESSIONNEL_BE]"
+    MARKET_SHARE: "[PART_DE_MARCHE]"
diff --git a/conf/default.yaml b/conf/default.yaml
deleted file mode 100644
index e9ab84c..0000000
--- a/conf/default.yaml
+++ /dev/null
@@ -1,227 +0,0 @@
-# =======================
-# CONFIGURATION PRESIDIO 
-# =======================
-supported_languages: [en, fr]
-
-nlp_configuration:
-  nlp_engine_name: spacy
-  models:
-    - lang_code: en
-      model_name: en_core_web_lg
-    - lang_code: fr
-      model_name: fr_core_news_lg 
-      ner_model_configuration:
-        labels_to_ignore:
-          - LOCATION
-          - MISC
-          - CARDINAL
-          - EVENT
-          - LANGUAGE
-          - LAW
-          - ORDINAL
-          - PERCENT
-          - PRODUCT
-          - QUANTITY
-          - WORK_OF_ART
-        confidence_thresholds:
-          DEFAULT_CONFIDENCE: 0.85
-          PERSON: 0.85
-          ORGANIZATION: 0.55
-
-recognizer_registry:
-  load_predefined_recognizers: true
-  recognizers:
-    - name: FlexibleDateRecognizer
-      supported_language: fr
-      supported_entity: FLEXIBLE_DATE
-      patterns:
-        - name: Date format JJ mois AAAA
-          regex: "\\b(0?[1-9]|[12][0-9]|3[01])\\s+(janvier|février|mars|avril|mai|juin|juillet|août|septembre|octobre|novembre|décembre)\\s+(19|20)\\d{2}\\b"
-          score: 1.0
-        - name: Date format JJ/MM/AAAA
-          regex: "\\b(0[1-9]|[12][0-9]|3[01])[-/.](0[1-9]|1[012])[-/.](19|20)\\d{2}\\b"
-          score: 1.0
-      context: ["date", "né le", "signé le", "incident du"]
-
-    - name: BelgianAddressRecognizer
-      supported_language: fr
-      supported_entity: BE_ADDRESS
-      patterns:
-        - name: Adresse Belge complète
-          regex: "\\b(?:\\d{1,4}[A-Za-z]?(?:\\s*,)?\\s+)?(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève)(?:\\s+(?:de|la|le|d'|des))?(?:\\s+[A-Z][a-zà-ÿ'-]+)+,?(?:\\s+\\d{1,4}[A-Za-z]?)?,\\s*\\d{4}\\s+[A-Za-zà-ÿ'-]+"
-          score: 1.0
-      context: ["demeurant", "adresse", "siège social", "bureaux situés"]
-
-    - name: BelgianPhoneRecognizer
-      supported_language: fr
-      supported_entity: BE_PHONE_NUMBER
-      patterns:
-        - name: Numéro téléphone Belge (fixe ou mobile)
-          regex: "\\b0[1-9](?:[./\\s]?\\d{2,3}){3}\\b"
-          score: 0.95
-      context: ["Tel", "Tél", "téléphone", "gsm", "mobile"]
-
-    - name: SmartOrganizationRecognizer
-      supported_language: fr
-      supported_entity: ORGANIZATION
-      patterns:
-        - name: Nom + Forme légale (DigitalConsult SPRL)
-          regex: "\\b([A-Z][a-zà-ÿ]+(?:\\s[A-Z][a-zà-ÿ]+)*)\\s+(SPRL|SRL|SA|SCS|SNC)\\b"
-          score: 0.9
-        - name: Forme légale + Nom (SPRL DigitalConsult)
-          regex: "\\b(SPRL|SRL|SA|SCS|SNC)\\s+([A-Z][a-zà-ÿ]+(?:\\s[A-Z][a-zà-ÿ]+)*)\\b"
-          score: 0.9
-      context: ["société", "entreprise", "gérant de la"]
-
-    - name: ProfessionalIdRecognizer
-      supported_language: fr
-      supported_entity: BE_PRO_ID
-      patterns:
-        - name: Numéro IEC
-          regex: "(n°\\sIEC:?|IEC:?)\\s*\\d{6}"
-          score: 1.0
-      context: ["expert-comptable"]
-
-    - name: BelgianEnterpriseRecognizer
-      supported_language: fr
-      supported_entity: BE_ENTERPRISE_NUMBER
-      patterns:
-        - name: Numéro BCE/TVA Belge (avec ou sans BE)
-          regex: "\\b(BE)?\\s?0?\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3}\\b"
-          score: 1.0
-      context: ["BCE", "TVA", "intracommunautaire"]
-
-    - name: EmailRecognizer
-      supported_language: fr
-      supported_entity: EMAIL_ADDRESS
-      patterns:
-        - name: Email Pattern
-          regex: "\\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\\.[A-Z|a-z]{2,}\\b"
-          score: 1.0
-      context: ["email", "courriel", "mail"]
-
-    - name: IbanRecognizer
-      supported_language: fr
-      supported_entity: IBAN
-      patterns:
-        - name: IBAN Pattern
-          regex: "\\b[A-Z]{2}[0-9]{2}(?:\\s[0-9]{4}){3}\\b"
-          score: 1.0
-      context: ["iban", "compte"]
-
-    - name: BelgianNRNRecognizer
-      supported_language: fr
-      supported_entity: BE_NATIONAL_REGISTER_NUMBER
-      patterns:
-        - name: NRN Pattern
-          regex: "\\b[0-9]{2}\\.[0-9]{2}\\.[0-9]{2}-[0-9]{3}\\.[0-9]{2}\\b"
-          score: 1.0
-      context: ["registre national"]
-
-    - name: FrenchINSEERecognizer
-      supported_language: fr
-      supported_entity: FR_SOCIAL_SECURITY_NUMBER
-      patterns:
-        - name: INSEE Pattern with flexible spaces
-          regex: "\\b[12]\\s*[0-9]{2}\\s*(?:0[1-9]|1[0-2])\\s*(?:2[ABab]|[0-9]{2})\\s*[0-9]{3}\\s*[0-9]{3}[\\s]?[0-9]{2}\\b"
-          score: 0.95
-      context: ["sécurité sociale", "insee", "nir"]
-
-    - name: IpAddressRecognizer
-      supported_language: fr
-      supported_entity: IP_ADDRESS
-      patterns:
-        - name: IPv4
-          regex: "\\b(?:(?:25[0-5]|2[0-4][0-9]|1[0-9]{2}|[1-9]?[0-9])\\.){3}(?:25[0-5]|2[0-4][0-9]|1[0-9]{2}|[1-9]?[0-9])\\b"
-          score: 1.0
-        - name: IPv6
-          regex: "\\b([0-9a-fA-F]{1,4}:){7}[0-9a-fA-F]{1,4}\\b"
-          score: 0.9
-
-
-allow_list:
-  - Adresse
-  - ADRESSE
-  - Contrat
-  - Document
-  - Société
-  - Investisseur
-  - Montant
-  - Prêt
-  - Intérêt
-  - Partie
-  - Parties
-  - Annexe
-  - Remboursement
-  - Conversion
-  - Financement
-  - Sortie
-  - "Juste Valeur Marchande"
-  - Échéance
-  - Clause
-  - Clauses
-  - Principe
-  - Coûts
-  - Notifications
-  - Article
-  - Paragraphe
-  - Directeur
-  - Gérant
-  - Président
-  - DocuSign
-  - SPRL
-  - SA
-  - Loi
-  - Code
-  - Règlement
-  - Décret
-  - Arrêté
-  - Euro
-  - EUR
-  - Euros
-  - Taux
-  - Valeur
-  - Prix
-  - Coordonnées
-  - Témoins
-  - "Coordonnées bancaires"
-  - "Témoins clés"
-  - "montrent"
-  - "montrent des"
-  - "montrent des irrégularités"
-  - "bénéficiaire"
-
-anonymizer_config:
-  default_anonymizers:
-    PERSON: replace
-    LOCATION: replace
-    ORGANIZATION: replace
-    DATE_TIME: replace
-    MONEY: replace
-    EMAIL_ADDRESS: replace
-    IBAN: replace
-    BE_ENTERPRISE_NUMBER: replace
-    BE_NATIONAL_REGISTER_NUMBER: replace
-    FR_SOCIAL_SECURITY_NUMBER: replace
-    BE_PHONE_NUMBER: replace
-    FLEXIBLE_DATE: replace
-    BE_ADDRESS: replace
-    BE_PRO_ID: replace
-    IP_ADDRESS: replace
-
-  replacements:
-    PERSON: "<PERSONNE>"
-    LOCATION: "<LIEU>"
-    ORGANIZATION: "<ORGANISATION>"
-    DATE_TIME: "<DATE>"
-    MONEY: "<MONTANT>"
-    EMAIL_ADDRESS: "<EMAIL>"
-    IBAN: "<IBAN>"
-    BE_ENTERPRISE_NUMBER: "<NUM_ENTREPRISE_BE>"
-    BE_NATIONAL_REGISTER_NUMBER: "<NRN_BELGE>"
-    FR_SOCIAL_SECURITY_NUMBER: "<NUM_SECU_FR>"
-    BE_PHONE_NUMBER: "<TELEPHONE_BE>"
-    FLEXIBLE_DATE: "<DATE>"
-    BE_ADDRESS: "<ADRESSE_BELGE>"
-    BE_PRO_ID: "<ID_PROFESSIONNEL>"
-    IP_ADDRESS: "<ADRESSE_IP>"
diff --git a/conf/main.yaml b/conf/main.yaml
new file mode 100644
index 0000000..586396c
--- /dev/null
+++ b/conf/main.yaml
@@ -0,0 +1,30 @@
+# =======================
+# CONFIGURATION PRESIDIO MODULAIRE
+# =======================
+
+# Langues supportées
+supported_languages: [en, fr]
+default_language: fr
+
+# Inclusion des modules de configuration
+includes:
+  # Configuration NLP (spaCy préservée)
+  - nlp/spacy_config.yaml
+
+  # Recognizers PII par dossier (garder uniquement les dossiers récents)
+  - recognizers/PII/belgian/*
+  - recognizers/PII/french/*
+  - recognizers/PII/generic/*
+
+  # Recognizers Business par dossier
+  - recognizers/Business/belgian/*
+  - recognizers/Business/french/*
+
+  # Configuration d'anonymisation
+  - anonymization/*
+
+# Configuration globale simplifiée
+global_settings:
+  version: "2.0.0"
+  cache_enabled: true
+  timeout_seconds: 30
diff --git a/conf/nlp/spacy_config.yaml b/conf/nlp/spacy_config.yaml
new file mode 100644
index 0000000..7def878
--- /dev/null
+++ b/conf/nlp/spacy_config.yaml
@@ -0,0 +1,33 @@
+nlp_configuration:
+  nlp_engine_name: spacy
+  models:
+    - lang_code: en
+      model_name: en_core_web_lg
+    - lang_code: fr
+      model_name: fr_core_news_lg
+
+  # Configuration NER globale (sans confidence_thresholds)
+  ner_model_configuration:
+    model_to_presidio_entity_mapping:
+      PER: PERSON
+      PERSON: PERSON
+      ORG: ORGANIZATION
+      ORGANIZATION: ORGANIZATION
+      LOC: LOCATION
+      LOCATION: LOCATION
+      DATE: DATE_TIME
+      TIME: DATE_TIME
+      MISC: DATE_TIME
+    labels_to_ignore:
+      - LOCATION
+      - MISC
+      - CARDINAL
+      - EVENT
+      - LANGUAGE
+      - LAW
+      - ORDINAL
+      - PERCENT
+      - PRODUCT
+      - QUANTITY
+      - WORK_OF_ART
+    low_score_entity_names: []
diff --git a/conf/recognizers/Business/belgian/enterprise_numbers.yaml b/conf/recognizers/Business/belgian/enterprise_numbers.yaml
new file mode 100644
index 0000000..b6da68c
--- /dev/null
+++ b/conf/recognizers/Business/belgian/enterprise_numbers.yaml
@@ -0,0 +1,24 @@
+# Recognizer pour numéros d'entreprise belges
+recognizer_registry:
+  recognizers:
+    - name: BelgianEnterpriseRecognizer
+      supported_language: fr
+      supported_entity: BE_ENTERPRISE_NUMBER
+      patterns:
+        - name: Numéro BCE avec deux points
+          regex: "(?<=\\bBCE\\s*:\\s*)((BE)?\\s?0\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3})\\b"
+          score: 1.0
+        - name: Numéro TVA avec deux points
+          regex: "(?<=\\bTVA\\s*:\\s*)(BE\\d{4}\\.\\d{3}\\.\\d{3})\\b"
+          score: 1.0
+        - name: Numéro d'entreprise général
+          regex: "(?<!(?:BCE|TVA)\\s*:\\s*)\\b(BE)?\\s?0\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3}\\b"
+          score: 0.9
+        - name: Numéro ONSS
+          regex: "\\bONSS\\s*:?\\s*\\d{7}\\b"
+          score: 0.95
+        - name: Numéro patronal
+          regex: "\\b(?:numéro\\s+)?patronal\\s*:?\\s*\\d{7}\\b"
+          score: 0.9
+      context:
+        ["TVA", "intracommunautaire", "ONSS", "entreprise", "patronal"]
diff --git a/conf/recognizers/Business/belgian/organization_names.yaml b/conf/recognizers/Business/belgian/organization_names.yaml
new file mode 100644
index 0000000..a907189
--- /dev/null
+++ b/conf/recognizers/Business/belgian/organization_names.yaml
@@ -0,0 +1,28 @@
+# Recognizer pour noms d'organisations belges
+recognizer_registry:
+  recognizers:
+    - name: SmartOrganizationRecognizer
+      supported_language: fr
+      supported_entity: ORGANIZATION
+      patterns:
+        # Noms avec suffixes typiques d'entreprise
+        - name: Noms entreprise avec suffixes
+          regex: "\\b([A-Z][a-zA-Zà-ÿ]+(?:Consult|Tech|Soft|Digital|Solutions|Services|Group|Corp|Company|Systems|Data|Cloud|Web|Net|Info|Cyber|Smart|Pro|Expert|Plus|Max|Global|International|Europe|Belgium|Brussels|Wallonie|Flandre))\\b(?!\\s*\\([^)]*(?:BCE|TVA)[^)]*\\))"
+          score: 0.9
+
+        # Formes légales complètes avec nom d'entreprise
+        - name: Formes légales complètes
+          regex: "\\b((?:SPRL|SRL|SA|ASBL|SCS|SNC)\\s+[A-Z][a-zA-Zà-ÿ]+(?:\\s+[A-Z][a-zA-Zà-ÿ]+)*)(?!\\s*\\([^)]*(?:BCE|TVA)[^)]*\\))"
+          score: 0.95
+
+        # Noms d'entreprise avec contexte spécifique (garder pour autres cas)
+        - name: Noms avec contexte entreprise
+          regex: "(?<=\\b(?:société|entreprise)\\s+)([A-Z][a-zA-Zà-ÿ]+(?:\\s+[A-Z][a-zA-Zà-ÿ]+)*)(?!\\s*\\([^)]*(?:BCE|TVA)[^)]*\\))"
+          score: 0.85
+
+        # Noms précédés de "gérant de la"
+        - name: Noms après gérant
+          regex: "(?<=gérant\\s+de\\s+la\\s+)([A-Z][a-zA-Zà-ÿ]+(?:\\s+[A-Z][a-zA-Zà-ÿ]+)*)(?!\\s*\\([^)]*(?:BCE|TVA)[^)]*\\))"
+          score: 0.8
+
+      context: ["société", "entreprise", "gérant de la", "administrateur"]
diff --git a/conf/recognizers/Business/belgian/professional_ids.yaml b/conf/recognizers/Business/belgian/professional_ids.yaml
new file mode 100644
index 0000000..74e124e
--- /dev/null
+++ b/conf/recognizers/Business/belgian/professional_ids.yaml
@@ -0,0 +1,20 @@
+# Recognizer pour identifiants professionnels belges
+recognizer_registry:
+  recognizers:
+    - name: BelgianProfessionalIdRecognizer
+      supported_language: fr
+      supported_entity: BE_PROFESSIONAL_ID
+      patterns:
+        - name: Numéro IEC avec deux points
+          regex: "(?<=\\bIEC\\s*:\\s*)\\d{6}\\b"
+          score: 1.0
+        - name: Numéro IEC général
+          regex: "(?<!IEC\\s*:\\s*)\\b(?:n°\\s*)?IEC\\s*:?\\s*\\d{6}\\b"
+          score: 0.9
+        - name: Numéro d'avocat
+          regex: "\\b(?:avocat\\s+)?n°\\s*\\d{4,6}\\b"
+          score: 0.8
+        - name: Numéro de médecin
+          regex: "\\b(?:Dr\\.|médecin)\\s*n°\\s*\\d{5,7}\\b"
+          score: 0.85
+      context: ["expert-comptable", "IEC", "avocat", "médecin", "professionnel"]
diff --git a/conf/recognizers/Business/french/siret_siren.yaml b/conf/recognizers/Business/french/siret_siren.yaml
new file mode 100644
index 0000000..0e2bb5e
--- /dev/null
+++ b/conf/recognizers/Business/french/siret_siren.yaml
@@ -0,0 +1,17 @@
+# Recognizer pour SIRET/SIREN français
+recognizer_registry:
+  recognizers:
+    - name: FrenchSIRETRecognizer
+      supported_language: fr
+      supported_entity: FR_SIRET
+      patterns:
+        - name: SIRET complet
+          regex: "\\b[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{5}\\b"
+          score: 0.9
+        - name: SIREN
+          regex: "\\b[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{3}\\b"
+          score: 0.85
+        - name: SIRET avec espaces
+          regex: "\\bSIRET\\s*:?\\s*[0-9]{3}\\s[0-9]{3}\\s[0-9]{3}\\s[0-9]{5}\\b"
+          score: 0.95
+      context: ["SIRET", "SIREN", "établissement", "entreprise", "société"]
diff --git a/conf/recognizers/Business/generic/market_share.yaml b/conf/recognizers/Business/generic/market_share.yaml
new file mode 100644
index 0000000..84bebc1
--- /dev/null
+++ b/conf/recognizers/Business/generic/market_share.yaml
@@ -0,0 +1,34 @@
+# Recognizer pour parts de marché
+recognizer_registry:
+  recognizers:
+    - name: MarketShareRecognizer
+      supported_language: fr
+      supported_entity: MARKET_SHARE
+      patterns:
+        # Pourcentages de marché
+        - name: Market Share Percentage
+          regex: "\\b(?:détient|possède|contrôle|représente)?\\s*(?:environ\\s+)?(?:\\d{1,2}(?:[,.]\\d{1,2})?%)\\s*(?:de\\s+(?:part\\s+de\\s+)?marché|du\\s+marché|de\\s+parts?)\\b"
+          score: 0.9
+        
+        # Positions de marché
+        - name: Market Position
+          regex: "\\b(?:leader|numéro\\s+\\d+|\\d+(?:er|ème)\\s+acteur|position\\s+dominante|monopole)\\s+(?:du\\s+)?(?:marché|secteur)\\b"
+          score: 0.85
+        
+        # Parts relatives
+        - name: Relative Market Share
+          regex: "\\b(?:majoritaire|minoritaire|principale|significative)\\s+(?:part\\s+de\\s+)?marché\\b"
+          score: 0.8
+        
+        # Données de concentration
+        - name: Market Concentration
+          regex: "\\b(?:concentration|consolidation|fusion)\\s+(?:du\\s+)?marché\\b"
+          score: 0.75
+        
+        # Chiffres d'affaires relatifs
+        - name: Revenue Share
+          regex: "\\b(?:\\d{1,2}(?:[,.]\\d{1,2})?%)\\s*(?:du\\s+)?(?:chiffre\\s+d'affaires|CA|revenus?)\\s+(?:du\\s+)?(?:marché|secteur)\\b"
+          score: 0.85
+      
+      context:
+        ["part de marché", "position concurrentielle", "leader", "concurrent", "secteur", "industrie", "chiffre d'affaires", "revenus", "concentration", "monopole", "oligopole"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/belgian/addresses.yaml b/conf/recognizers/PII/belgian/addresses.yaml
new file mode 100644
index 0000000..d17fdee
--- /dev/null
+++ b/conf/recognizers/PII/belgian/addresses.yaml
@@ -0,0 +1,44 @@
+# Recognizer pour adresses belges complètes
+recognizer_registry:
+  recognizers:
+    - name: BelgianAddressRecognizer
+      supported_language: fr
+      supported_entity: BE_ADDRESS
+      patterns:
+        # Pattern principal : numéro + rue + code postal + ville (SANS contexte)
+        - name: Adresse complète avec numéro devant
+          regex: "\\b\\d{1,4}[A-Za-z]?\\s+(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\\s+(?:de\\s+la\\s+|de\\s+|du\\s+|des\\s+|d'|la\\s+|le\\s+)?[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*\\s*,\\s*[1-9]\\d{3}\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*"
+          score: 1.0
+
+        # Pattern avec lookbehind positif pour exclure "demeurant" de la capture
+        - name: Adresse après contexte demeurant
+          regex: "(?<=\\bdemeurant\\s)\\d{1,4}[A-Za-z]?\\s+(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\\s+(?:de\\s+la\\s+|de\\s+|du\\s+|des\\s+|d'|la\\s+|le\\s+)?[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*\\s*,\\s*[1-9]\\d{3}\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*"
+          score: 1.0
+
+        # Pattern avec lookbehind pour autres contextes
+        - name: Adresse après contexte résidant
+          regex: "(?<=\\b(?:résidant|domicilié|habite|situé)\\s)\\d{1,4}[A-Za-z]?\\s+(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\\s+(?:de\\s+la\\s+|de\\s+|du\\s+|des\\s+|d'|la\\s+|le\\s+)?[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*\\s*,\\s*[1-9]\\d{3}\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*"
+          score: 1.0
+
+        # Pattern avec lookbehind pour contexte bureaux
+        - name: Adresse après contexte bureaux
+          regex: "(?<=\\b(?:dans les bureaux situés|siège social situé)\\s)\\d{1,4}[A-Za-z]?\\s+(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\\s+(?:de\\s+la\\s+|de\\s+|du\\s+|des\\s+|d'|la\\s+|le\\s+)?[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*\\s*,\\s*[1-9]\\d{3}\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*"
+          score: 0.9
+
+        # Pattern alternatif : rue + numéro + ville (format classique)
+        - name: Adresse format classique
+          regex: "\\b(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\\s+(?:de\\s+la\\s+|de\\s+|du\\s+|des\\s+|d'|la\\s+|le\\s+)?[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*(?:\\s*,?\\s*\\d{1,4}[A-Za-z]?)?\\s*,\\s*[1-9]\\d{3}\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+(?:\\s+[A-ZÀ-Ÿ][a-zà-ÿ'-]+)*"
+          score: 0.8
+
+      context:
+        [
+          "demeurant",
+          "résidant",
+          "domicilié",
+          "habite",
+          "situé au",
+          "sis à",
+          "dans les bureaux situés",
+          "siège social",
+          "adresse",
+        ]
diff --git a/conf/recognizers/PII/belgian/documents.yaml b/conf/recognizers/PII/belgian/documents.yaml
new file mode 100644
index 0000000..1483fd6
--- /dev/null
+++ b/conf/recognizers/PII/belgian/documents.yaml
@@ -0,0 +1,26 @@
+# Recognizer pour documents belges
+recognizer_registry:
+  recognizers:
+    - name: BelgianIDCardRecognizer
+      supported_language: fr
+      supported_entity: BE_ID_CARD
+      patterns:
+        - name: Carte d'identité belge
+          regex: "\\b[0-9]{3}\\-[0-9]{7}\\-[0-9]{2}\\b"
+          score: 0.95
+        - name: eID compact
+          regex: "\\b[0-9]{12}\\b"
+          score: 0.8
+      context: ["carte d'identité", "eID", "identiteitskaart", "pièce d'identité"]
+
+    - name: BelgianPassportRecognizer
+      supported_language: fr
+      supported_entity: BE_PASSPORT
+      patterns:
+        - name: Passeport belge
+          regex: "\\b[A-Z]{2}[0-9]{6}\\b"
+          score: 0.95
+        - name: Passeport avec espaces
+          regex: "\\b[A-Z]{2}\\s[0-9]{6}\\b"
+          score: 0.9
+      context: ["passeport", "passport", "paspoort", "document de voyage"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/belgian/national_register.yaml b/conf/recognizers/PII/belgian/national_register.yaml
new file mode 100644
index 0000000..34b8f95
--- /dev/null
+++ b/conf/recognizers/PII/belgian/national_register.yaml
@@ -0,0 +1,25 @@
+# Recognizer pour numéro de registre national belge
+recognizer_registry:
+  recognizers:
+    - name: BelgianNRNRecognizer
+      supported_language: fr
+      supported_entity: BE_NATIONAL_REGISTER_NUMBER
+      patterns:
+        - name: NRN avec points et tiret
+          regex: "\\b[0-9]{2}\\.[0-9]{2}\\.[0-9]{2}-[0-9]{3}\\.[0-9]{2}\\b"
+          score: 1.0
+        - name: NRN compact
+          regex: "\\b[0-9]{11}\\b"
+          score: 0.7
+        - name: NRN avec espaces
+          regex: "\\b[0-9]{2}\\s[0-9]{2}\\s[0-9]{2}\\s[0-9]{3}\\s[0-9]{2}\\b"
+          score: 0.8
+      context:
+        [
+          "registre national",
+          "numéro national",
+          "NN",
+          "RN",
+          "identité",
+          "carte d'identité",
+        ]
diff --git a/conf/recognizers/PII/belgian/phones.yaml b/conf/recognizers/PII/belgian/phones.yaml
new file mode 100644
index 0000000..eda9bc2
--- /dev/null
+++ b/conf/recognizers/PII/belgian/phones.yaml
@@ -0,0 +1,44 @@
+# Recognizer pour numéros de téléphone belges
+recognizer_registry:
+  recognizers:
+    - name: BelgianPhoneRecognizer
+      supported_language: fr
+      supported_entity: BE_PHONE_NUMBER
+      patterns:
+        # Patterns avec contexte Tel: et Tél:
+        - name: Téléphone fixe avec contexte Tel
+          regex: '(?<=Tel\s*:\s*)0[1-9](?:[./\s]?\d{2,3}){3}(?=\s|\)|$|[.,;])'
+          score: 0.99
+        - name: Téléphone fixe avec contexte Tél
+          regex: '(?<=Tél\s*:\s*)0[1-9](?:[./\s]?\d{2,3}){3}(?=\s|\)|$|[.,;])'
+          score: 0.99
+        - name: Mobile avec contexte Tel
+          regex: '(?<=Tel\s*:\s*)04[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}(?=\s|\)|$|[.,;])'
+          score: 0.99
+        - name: Mobile avec contexte Tél
+          regex: '(?<=Tél\s*:\s*)04[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}(?=\s|\)|$|[.,;])'
+          score: 0.99
+        # Patterns généraux (sans contexte spécifique)
+        - name: Téléphone fixe belge
+          regex: '(?<!BCE\s*:?\s*)\b0[1-9](?:[./\s]?\d{2,3}){3}(?=\b|\)|$|[.,;])(?!.*BCE)'
+          score: 0.95
+        - name: Mobile belge
+          regex: '\b04[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}[./\s]?[0-9]{2}(?=\b|\)|$|[.,;])'
+          score: 0.98
+        - name: International belge
+          regex: '\+32\s?[1-9](?:[./\s]?\d{2,3}){3}(?=\b|\)|$|[.,;])'
+          score: 0.99
+        - name: Numéro vert belge
+          regex: '\b0800[./\s]?[0-9]{2}[./\s]?[0-9]{3}(?=\b|\)|$|[.,;])'
+          score: 0.9
+      context:
+        [
+          "Tel",
+          "Tél",
+          "téléphone",
+          "gsm",
+          "mobile",
+          "portable",
+          "appeler",
+          "joindre",
+        ]
diff --git a/conf/recognizers/PII/french/addresses.yaml b/conf/recognizers/PII/french/addresses.yaml
new file mode 100644
index 0000000..0d512df
--- /dev/null
+++ b/conf/recognizers/PII/french/addresses.yaml
@@ -0,0 +1,14 @@
+# Recognizer pour adresses françaises
+recognizer_registry:
+  recognizers:
+    - name: FrenchAddressRecognizer
+      supported_language: fr
+      supported_entity: FR_ADDRESS
+      patterns:
+        - name: Adresse française complète
+          regex: "\\b\\d{1,4}(?:bis|ter|quater)?\\s+(?:rue|avenue|boulevard|place|impasse|allée|chemin|route)\\s+[A-Za-zà-ÿ\\s'-]+,\\s*[0-9]{5}\\s+[A-Za-zà-ÿ\\s'-]+\\b"
+          score: 0.95
+        - name: Code postal français
+          regex: "\\b[0-9]{5}\\b"
+          score: 0.6
+      context: ["adresse", "domicile", "résidence", "siège social"]
diff --git a/conf/recognizers/PII/french/documents.yaml b/conf/recognizers/PII/french/documents.yaml
new file mode 100644
index 0000000..ee4bccf
--- /dev/null
+++ b/conf/recognizers/PII/french/documents.yaml
@@ -0,0 +1,39 @@
+# Recognizer pour documents d'identité français
+recognizer_registry:
+  recognizers:
+    - name: FrenchIDCardRecognizer
+      supported_language: fr
+      supported_entity: FR_ID_CARD
+      patterns:
+        - name: Numéro CNI nouveau format
+          regex: "\\b[0-9]{12}\\b"
+          score: 0.85
+        - name: Numéro CNI avec espaces
+          regex: "\\b[0-9]{4}\\s[0-9]{4}\\s[0-9]{4}\\b"
+          score: 0.9
+      context:
+        ["carte d'identité", "CNI", "pièce d'identité", "numéro d'identité"]
+
+    - name: FrenchPassportRecognizer
+      supported_language: fr
+      supported_entity: FR_PASSPORT
+      patterns:
+        - name: Numéro de passeport français
+          regex: "\\b[0-9]{2}[A-Z]{2}[0-9]{5}\\b"
+          score: 0.95
+        - name: Passeport avec espaces
+          regex: "\\b[0-9]{2}\\s[A-Z]{2}\\s[0-9]{5}\\b"
+          score: 0.9
+      context: ["passeport", "passport", "document de voyage"]
+
+    - name: FrenchDriverLicenseRecognizer
+      supported_language: fr
+      supported_entity: FR_DRIVER_LICENSE
+      patterns:
+        - name: Permis de conduire français
+          regex: "\\b[0-9]{12}\\b"
+          score: 0.8
+        - name: Permis avec format
+          regex: "\\b(?:permis\\s+(?:de\\s+)?conduire\\s*:?\\s*)?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{4}\\b"
+          score: 0.9
+      context: ["permis de conduire", "permis", "conduire", "licence"]
diff --git a/conf/recognizers/PII/french/financial.yaml b/conf/recognizers/PII/french/financial.yaml
new file mode 100644
index 0000000..39f8b3e
--- /dev/null
+++ b/conf/recognizers/PII/french/financial.yaml
@@ -0,0 +1,26 @@
+# Recognizer pour données financières françaises
+recognizer_registry:
+  recognizers:
+    - name: FrenchTaxIDRecognizer
+      supported_language: fr
+      supported_entity: FR_TAX_ID
+      patterns:
+        - name: Numéro fiscal français
+          regex: "\\b[0-9]{13}\\b"
+          score: 0.85
+        - name: Référence fiscale
+          regex: "\\b(?:numéro\\s+fiscal\\s*:?\\s*)?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{5}\\b"
+          score: 0.9
+      context: ["numéro fiscal", "référence fiscale", "impôts", "SIP"]
+
+    - name: FrenchBankAccountRecognizer
+      supported_language: fr
+      supported_entity: FR_BANK_ACCOUNT
+      patterns:
+        - name: RIB français
+          regex: "\\b[0-9]{5}\\s[0-9]{5}\\s[0-9]{11}\\s[0-9]{2}\\b"
+          score: 0.95
+        - name: Numéro de compte
+          regex: "\\b(?:compte\\s*:?\\s*)?[0-9]{5}[0-9]{5}[0-9]{11}[0-9]{2}\\b"
+          score: 0.9
+      context: ["RIB", "compte bancaire", "numéro de compte", "relevé"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/french/social_security.yaml b/conf/recognizers/PII/french/social_security.yaml
new file mode 100644
index 0000000..71f323c
--- /dev/null
+++ b/conf/recognizers/PII/french/social_security.yaml
@@ -0,0 +1,18 @@
+# Recognizer pour numéros INSEE français
+recognizer_registry:
+  recognizers:
+    - name: FrenchINSEERecognizer
+      supported_language: fr
+      supported_entity: FR_SOCIAL_SECURITY_NUMBER
+      patterns:
+        - name: INSEE complet avec espaces
+          regex: "\\b[12]\\s*[0-9]{2}\\s*(?:0[1-9]|1[0-2])\\s*(?:2[ABab]|[0-9]{2})\\s*[0-9]{3}\\s*[0-9]{3}[\\s]?[0-9]{2}\\b"
+          score: 0.95
+        - name: NIR compact
+          regex: "\\b[12][0-9]{12}[0-9]{2}\\b"
+          score: 0.85
+        - name: INSEE avec tirets
+          regex: "\\b[12]-[0-9]{2}-[0-9]{2}-[0-9]{2}-[0-9]{3}-[0-9]{3}-[0-9]{2}\\b"
+          score: 0.9
+      context:
+        ["sécurité sociale", "insee", "nir", "numéro de sécu", "carte vitale"]
diff --git a/conf/recognizers/PII/generic/bank_accounts.yaml b/conf/recognizers/PII/generic/bank_accounts.yaml
new file mode 100644
index 0000000..5fd0c1b
--- /dev/null
+++ b/conf/recognizers/PII/generic/bank_accounts.yaml
@@ -0,0 +1,20 @@
+# Recognizer pour IBAN
+recognizer_registry:
+  recognizers:
+    - name: IbanRecognizer
+      supported_language: fr
+      supported_entity: IBAN
+      patterns:
+        - name: IBAN avec espaces
+          regex: "\\b[A-Z]{2}[0-9]{2}(?:\\s[0-9A-Z]{4}){3,7}\\b"
+          score: 1.0
+        - name: IBAN compact
+          regex: "\\b[A-Z]{2}[0-9]{2}[0-9A-Z]{12,30}\\b"
+          score: 0.9
+        - name: IBAN belge spécifique
+          regex: "\\bBE[0-9]{2}\\s?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{4}\\b"
+          score: 0.95
+        - name: IBAN français spécifique
+          regex: "\\bFR[0-9]{2}\\s?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{4}\\s?[0-9]{3}\\b"
+          score: 0.95
+      context: ["iban", "compte", "bancaire", "virement", "RIB"]
diff --git a/conf/recognizers/PII/generic/biometric_data.yaml b/conf/recognizers/PII/generic/biometric_data.yaml
new file mode 100644
index 0000000..7d28125
--- /dev/null
+++ b/conf/recognizers/PII/generic/biometric_data.yaml
@@ -0,0 +1,14 @@
+# Recognizer pour données biométriques
+recognizer_registry:
+  recognizers:
+    - name: BiometricDataRecognizer
+      supported_language: fr
+      supported_entity: BIOMETRIC_DATA
+      patterns:
+        - name: Données biométriques
+          regex: "\\b(?:empreinte(?:s)?\\s+digitale(?:s)?|reconnaissance\\s+faciale|scan\\s+(?:iris|rétine)|biométrie|ADN|profil\\s+génétique)\\b"
+          score: 0.95
+        - name: Identifiants biométriques
+          regex: "\\b(?:template|hash)\\s+biométrique\\s*:?\\s*[A-F0-9]{32,}\\b"
+          score: 0.9
+      context: ["biométrie", "empreinte", "reconnaissance", "scan", "identification"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/generic/dates.yaml b/conf/recognizers/PII/generic/dates.yaml
new file mode 100644
index 0000000..92764ae
--- /dev/null
+++ b/conf/recognizers/PII/generic/dates.yaml
@@ -0,0 +1,74 @@
+recognizer_registry:
+  recognizers:
+    - name: DateTimeRecognizer
+      supported_language: fr
+      supported_entity: DATE_TIME
+      patterns:
+        # Formats français standards avec différents séparateurs
+        - name: Date française DD/MM/YYYY
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])/(?:0?[1-9]|1[0-2])/(?:19|20)\\d{2}\\b"
+          score: 0.95
+
+        - name: Date française DD-MM-YYYY
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])-(?:0?[1-9]|1[0-2])-(?:19|20)\\d{2}\\b"
+          score: 0.95
+
+        - name: Date française DD MM YYYY (espaces)
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])\\s+(?:0?[1-9]|1[0-2])\\s+(?:19|20)\\d{2}\\b"
+          score: 0.9
+
+        - name: Date ISO YYYY-MM-DD
+          regex: "\\b(?:19|20)\\d{2}-(?:0[1-9]|1[0-2])-(?:0[1-9]|[12][0-9]|3[01])\\b"
+          score: 0.98
+
+        # Dates avec mois en lettres (joli format)
+        - name: Date avec mois en lettres
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])\\s+(?:janvier|février|mars|avril|mai|juin|juillet|août|septembre|octobre|novembre|décembre)\\s+(?:19|20)\\d{2}\\b"
+          score: 0.99
+
+        # Format belge DD.MM.YYYY
+        - name: Date belge DD.MM.YYYY
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])\\.(?:0?[1-9]|1[0-2])\\.(?:19|20)\\d{2}\\b"
+          score: 0.95
+
+        # Heures (HH:MM et HH:MM:SS)
+        - name: Heure HH:MM(:SS)?
+          regex: "\\b(?:[01]?[0-9]|2[0-3]):[0-5][0-9](?::[0-5][0-9])?\\b"
+          score: 0.85
+
+        # Date et heure combinées (ex: 12/05/2023 14:30)
+        - name: Date et heure combinées
+          regex: "\\b(?:0?[1-9]|[12][0-9]|3[01])[/-](?:0?[1-9]|1[0-2])[/-](?:19|20)\\d{2}\\s+(?:[01]?[0-9]|2[0-3]):[0-5][0-9](?::[0-5][0-9])?\\b"
+          score: 0.97
+
+        # Années seules dans un contexte fort
+        - name: Année avec contexte
+          regex: "\\b(?:en|depuis|année|an|né en|décédé en)\\s+(?:19|20)\\d{2}\\b"
+          score: 0.8
+
+      context:
+        [
+          "date",
+          "né le",
+          "née le",
+          "naissance",
+          "décès",
+          "décédé le",
+          "le",
+          "du",
+          "au",
+          "depuis",
+          "jusqu'au",
+          "entre",
+          "avant",
+          "après",
+          "heure",
+          "horaire",
+          "rendez-vous",
+          "réunion",
+          "événement",
+          "signature",
+          "signé le",
+          "établi le",
+          "fait le",
+        ]
diff --git a/conf/recognizers/PII/generic/emails.yaml b/conf/recognizers/PII/generic/emails.yaml
new file mode 100644
index 0000000..aed2bdd
--- /dev/null
+++ b/conf/recognizers/PII/generic/emails.yaml
@@ -0,0 +1,18 @@
+# Recognizer pour adresses email
+recognizer_registry:
+  recognizers:
+    - name: EmailRecognizer
+      supported_language: fr
+      supported_entity: EMAIL_ADDRESS
+      patterns:
+        - name: Email standard
+          regex: "\\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\\.[A-Z|a-z]{2,}\\b"
+          score: 1.0
+        - name: Email obfusqué
+          regex: "\\b[A-Za-z0-9._%+-]+\\s*\\[at\\]\\s*[A-Za-z0-9.-]+\\s*\\[dot\\]\\s*[A-Z|a-z]{2,}\\b"
+          score: 0.8
+        - name: Email avec (at) et (point)
+          regex: "\\b[A-Za-z0-9._%+-]+\\s*\\(at\\)\\s*[A-Za-z0-9.-]+\\s*\\(point\\)\\s*[A-Z|a-z]{2,}\\b"
+          score: 0.7
+      context:
+        ["email", "courriel", "mail", "@", "contact", "adresse électronique"]
diff --git a/conf/recognizers/PII/generic/financial_data.yaml b/conf/recognizers/PII/generic/financial_data.yaml
new file mode 100644
index 0000000..47c99bd
--- /dev/null
+++ b/conf/recognizers/PII/generic/financial_data.yaml
@@ -0,0 +1,17 @@
+# Recognizer pour données financières RGPD
+recognizer_registry:
+  recognizers:
+    - name: RGPDFinancialDataRecognizer
+      supported_language: fr
+      supported_entity: RGPD_FINANCIAL_DATA
+      patterns:
+        - name: Numéro de carte bancaire
+          regex: "\\b(?:4[0-9]{12}(?:[0-9]{3})?|5[1-5][0-9]{14}|3[47][0-9]{13})\\b"
+          score: 1.0
+        - name: Code de sécurité
+          regex: "\\b(?:CVV|CVC|cryptogramme)\\s*:?\\s*[0-9]{3,4}\\b"
+          score: 0.95
+        - name: Revenus
+          regex: "\\b(?:salaire|revenu|rémunération)\\s*:?\\s*[0-9]{1,3}(?:[\\s.,][0-9]{3})*\\s*€?\\b"
+          score: 0.8
+      context: ["financier", "bancaire", "carte", "paiement", "salaire"]
diff --git a/conf/recognizers/PII/generic/health_data.yaml b/conf/recognizers/PII/generic/health_data.yaml
new file mode 100644
index 0000000..20032f1
--- /dev/null
+++ b/conf/recognizers/PII/generic/health_data.yaml
@@ -0,0 +1,17 @@
+# Recognizer pour données de santé
+recognizer_registry:
+  recognizers:
+    - name: HealthDataRecognizer
+      supported_language: fr
+      supported_entity: HEALTH_DATA
+      patterns:
+        - name: Informations médicales
+          regex: "\\b(?:maladie|pathologie|diagnostic|traitement|médicament|hospitalisation|chirurgie|opération|allergie|antécédent|symptôme)\\b"
+          score: 0.85
+        - name: Données médicales sensibles
+          regex: "\\b(?:VIH|SIDA|cancer|diabète|dépression|schizophrénie|bipolarité|addiction)\\b"
+          score: 0.95
+        - name: Professionnels de santé
+          regex: "\\b(?:Dr|Docteur|Médecin|Infirmier|Psychiatre|Psychologue)\\s+[A-Z][a-z]+\\b"
+          score: 0.9
+      context: ["santé", "médical", "hôpital", "clinique", "patient", "dossier médical"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/generic/ip_addresses.yaml b/conf/recognizers/PII/generic/ip_addresses.yaml
new file mode 100644
index 0000000..48896b6
--- /dev/null
+++ b/conf/recognizers/PII/generic/ip_addresses.yaml
@@ -0,0 +1,17 @@
+# Recognizer pour adresses IP
+recognizer_registry:
+  recognizers:
+    - name: IpAddressRecognizer
+      supported_language: fr
+      supported_entity: IP_ADDRESS
+      patterns:
+        - name: IPv4
+          regex: "\\b(?:(?:25[0-5]|2[0-4][0-9]|1[0-9]{2}|[1-9]?[0-9])\\.){3}(?:25[0-5]|2[0-4][0-9]|1[0-9]{2}|[1-9]?[0-9])\\b"
+          score: 1.0
+        - name: IPv6
+          regex: "\\b([0-9a-fA-F]{1,4}:){7}[0-9a-fA-F]{1,4}\\b"
+          score: 0.9
+        - name: IPv6 compressé
+          regex: "\\b([0-9a-fA-F]{1,4}:){1,7}:([0-9a-fA-F]{1,4}:){0,6}[0-9a-fA-F]{1,4}\\b"
+          score: 0.85
+      context: ["IP", "adresse", "serveur", "réseau", "connexion"]
diff --git a/conf/recognizers/PII/generic/political_opinions.yaml b/conf/recognizers/PII/generic/political_opinions.yaml
new file mode 100644
index 0000000..acfed90
--- /dev/null
+++ b/conf/recognizers/PII/generic/political_opinions.yaml
@@ -0,0 +1,11 @@
+# Recognizer pour opinions politiques
+recognizer_registry:
+  recognizers:
+    - name: PoliticalOpinionsRecognizer
+      supported_language: fr
+      supported_entity: POLITICAL_OPINIONS
+      patterns:
+        - name: Opinions politiques
+          regex: "\\b(?:vote|électeur|parti\\s+politique|opinion\\s+politique)\\b"
+          score: 0.7
+      context: ["données sensibles", "RGPD", "politique", "privé"]
\ No newline at end of file
diff --git a/conf/recognizers/PII/generic/sexual_orientation.yaml b/conf/recognizers/PII/generic/sexual_orientation.yaml
new file mode 100644
index 0000000..0fead2d
--- /dev/null
+++ b/conf/recognizers/PII/generic/sexual_orientation.yaml
@@ -0,0 +1,11 @@
+# Recognizer pour orientation sexuelle
+recognizer_registry:
+  recognizers:
+    - name: SexualOrientationRecognizer
+      supported_language: fr
+      supported_entity: SEXUAL_ORIENTATION
+      patterns:
+        - name: Orientation sexuelle
+          regex: "\\b(?:homosexuel|hétérosexuel|bisexuel|orientation\\s+sexuelle)\\b"
+          score: 0.9
+      context: ["données sensibles", "RGPD", "orientation", "privé"]
\ No newline at end of file
diff --git a/config_loader.py b/config_loader.py
new file mode 100644
index 0000000..18403ad
--- /dev/null
+++ b/config_loader.py
@@ -0,0 +1,107 @@
+import os
+import yaml
+import glob
+from typing import Dict, Any, List
+import logging
+
+logger = logging.getLogger(__name__)
+
+class ConfigLoader:
+    def __init__(self, config_dir: str = "conf"):
+        self.config_dir = config_dir
+        self.config = {}
+
+    def load_config(self, main_config_file: str = "main.yaml") -> Dict[str, Any]:
+        main_config_path = os.path.join(self.config_dir, main_config_file)
+
+        if not os.path.exists(main_config_path):
+            logger.warning(f"Fichier de configuration principal non trouvé: {main_config_path}")
+            return self._load_legacy_config()
+
+        with open(main_config_path, 'r', encoding='utf-8') as f:
+            main_config = yaml.safe_load(f)
+
+        if 'includes' in main_config:
+            for include_pattern in main_config['includes']:
+                self._load_includes(include_pattern)
+
+        self._merge_config(main_config)
+
+        logger.info(f"Configuration chargée avec {len(self.config.get('recognizer_registry', {}).get('recognizers', []))} recognizers")
+        return self.config
+
+    def _load_includes(self, pattern: str):
+        pattern = os.path.expandvars(pattern)
+        full_pattern = os.path.join(self.config_dir, pattern)
+        matching_files = glob.glob(full_pattern, recursive=True)
+
+        for file_path in sorted(matching_files):
+            if os.path.isfile(file_path) and file_path.endswith('.yaml'):
+                try:
+                    with open(file_path, 'r', encoding='utf-8') as f:
+                        module_config = yaml.safe_load(f)
+                        if module_config:
+                            self._merge_config(module_config)
+                            logger.debug(f"Module chargé: {file_path}")
+                except Exception as e:
+                    logger.error(f"Erreur lors du chargement de {file_path}: {e}")
+
+    def _merge_config(self, new_config: Dict[str, Any]):
+        for key, value in new_config.items():
+            if key == 'recognizer_registry':
+                if 'recognizer_registry' not in self.config:
+                    self.config['recognizer_registry'] = {'recognizers': []}
+
+                if 'recognizers' in value:
+                    self.config['recognizer_registry']['recognizers'].extend(value['recognizers'])
+
+                for reg_key, reg_value in value.items():
+                    if reg_key != 'recognizers':
+                        self.config['recognizer_registry'][reg_key] = reg_value
+
+            elif key == 'allow_list':
+                if 'allow_list' not in self.config:
+                    self.config['allow_list'] = []
+                if isinstance(value, list):
+                    self.config['allow_list'].extend(value)
+
+            elif key == 'nlp_configuration':
+                logger.info(f"🔧 Fusion de nlp_configuration: {value}")
+                if 'nlp_configuration' not in self.config:
+                    self.config['nlp_configuration'] = {}
+                self._merge_dict(self.config['nlp_configuration'], value)
+
+            elif isinstance(value, dict) and key in self.config and isinstance(self.config[key], dict):
+                self._merge_dict(self.config[key], value)
+            else:
+                self.config[key] = value
+
+    def _merge_dict(self, target: Dict[str, Any], source: Dict[str, Any]):
+        for key, value in source.items():
+            if isinstance(value, dict) and key in target and isinstance(target[key], dict):
+                self._merge_dict(target[key], value)
+            else:
+                target[key] = value
+
+    def _load_legacy_config(self) -> Dict[str, Any]:
+        legacy_path = os.path.join(self.config_dir, "default.yaml")
+        if os.path.exists(legacy_path):
+            logger.info("Utilisation de la configuration legacy: default.yaml")
+            with open(legacy_path, 'r', encoding='utf-8') as f:
+                return yaml.safe_load(f)
+        else:
+            raise FileNotFoundError(f"Aucun fichier de configuration trouvé dans {self.config_dir}")
+
+    def get_recognizers(self) -> List[Dict[str, Any]]:
+        return self.config.get('recognizer_registry', {}).get('recognizers', [])
+
+    def get_supported_languages(self) -> List[str]:
+        return self.config.get('supported_languages', ['fr'])
+
+    def load_single_file(self, file_path: str) -> Dict[str, Any]:
+        full_path = os.path.join(self.config_dir, file_path) if not os.path.isabs(file_path) else file_path
+        if not os.path.exists(full_path):
+            raise FileNotFoundError(f"Fichier de configuration non trouvé: {full_path}")
+
+        with open(full_path, 'r', encoding='utf-8') as f:
+            return yaml.safe_load(f)
diff --git a/docker-compose.yml b/docker-compose.yml
index 43761b0..ca96320 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -1,5 +1,3 @@
-version: '3.8'
-
 services:
   presidio-analyzer:
     build:
@@ -8,11 +6,4 @@ services:
     container_name: presidio-analyzer
     restart: unless-stopped
     ports:
-      - "5001"  # Port corrigé selon la doc Microsoft
-
-  presidio-anonymizer:
-    image: mcr.microsoft.com/presidio-anonymizer:latest
-    container_name: presidio-anonymizer
-    restart: unless-stopped
-    ports:
-      - "5002"  # Port corrigé selon la doc Microsoft
+      - "5001:5001"
diff --git a/entity_refiners.py b/entity_refiners.py
new file mode 100644
index 0000000..721376a
--- /dev/null
+++ b/entity_refiners.py
@@ -0,0 +1,56 @@
+from abc import ABC, abstractmethod
+from typing import Optional, Tuple
+import re
+import logging
+
+# Imports des raffineurs modulaires
+from refiners.iban_refiner import IBANRefiner
+from refiners.ip_refiner import IPAddressRefiner
+from refiners.date_refiner import DateRefiner
+from refiners.location_address_refiner import LocationAddressRefiner
+from refiners.word_boundary_refiner import WordBoundaryRefiner
+
+logger = logging.getLogger(__name__)
+
+class EntityRefiner(ABC):
+    """Classe de base pour le recadrage d'entités"""
+    
+    def __init__(self, entity_type: str):
+        self.entity_type = entity_type
+    
+    @abstractmethod
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Recadre une entité détectée"""
+        pass
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Vérifie si ce raffineur doit traiter ce type d'entité"""
+        return entity_type == self.entity_type
+
+class EntityRefinerManager:
+    """Gestionnaire des raffineurs d'entités"""
+    
+    def __init__(self):
+        self.refiners = [
+            WordBoundaryRefiner(),  # En premier pour étendre aux mots complets
+            IBANRefiner(),
+            IPAddressRefiner(),
+            DateRefiner(),
+            LocationAddressRefiner()
+        ]
+        logger.info(f"Initialized {len(self.refiners)} entity refiners")
+    
+    def register_refiner(self, refiner):
+        """Enregistre un nouveau raffineur"""
+        self.refiners.append(refiner)
+    
+    def refine_entity(self, text: str, entity_type: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Applique tous les raffineurs applicables à une entité"""
+        for refiner in self.refiners:
+            if refiner.should_process(entity_type):
+                result = refiner.refine(text, start, end)
+                if result:
+                    logger.debug(f"Entity refined by {refiner.__class__.__name__}: {start}-{end} -> {result[0]}-{result[1]}")
+                    return result
+        
+        return (start, end)
\ No newline at end of file
diff --git a/pipeline_manager.py b/pipeline_manager.py
new file mode 100644
index 0000000..75fdcda
--- /dev/null
+++ b/pipeline_manager.py
@@ -0,0 +1,68 @@
+from typing import List
+from presidio_analyzer import RecognizerResult
+from entity_refiners import EntityRefinerManager
+from post_processors import DeduplicationProcessor, OverlapResolver
+import logging
+
+logger = logging.getLogger(__name__)
+
+class AnalysisPipeline:
+    def __init__(self):
+        self.refiner_manager = EntityRefinerManager()
+        self.overlap_resolver = OverlapResolver()
+        self.deduplicator = DeduplicationProcessor()
+        logger.info("🚀 Pipeline d'analyse initialisé")
+    
+    def process(self, text: str, results: List[RecognizerResult], allow_list_terms: List[str]) -> List[RecognizerResult]:
+        """Traite les résultats à travers le pipeline complet"""
+        # 1. Filtrage allow-list
+        filtered_results = self._filter_allow_list(results, allow_list_terms, text)
+        
+        # 2. Raffinement individuel des entités
+        refined_results = []
+        for result in filtered_results:
+            refined_coords = self.refiner_manager.refine_entity(
+                text, 
+                result.entity_type, 
+                result.start, 
+                result.end
+            )
+            
+            if refined_coords is not None:
+                # Créer un nouveau RecognizerResult avec les coordonnées raffinées
+                refined_result = RecognizerResult(
+                    entity_type=result.entity_type,
+                    start=refined_coords[0],
+                    end=refined_coords[1],
+                    score=result.score
+                )
+                refined_results.append(refined_result)
+        
+        # 3. Résolution des chevauchements
+        resolved_results = self.overlap_resolver.process(refined_results, text)
+        
+        # 4. Déduplication
+        final_results = self.deduplicator.process(resolved_results, text)
+        
+        logger.info(f"🎯 Pipeline complet: {len(results)} -> {len(final_results)} entités")
+        return final_results
+    
+    def _filter_allow_list(self, results: List[RecognizerResult], allow_list_terms: List[str], text: str) -> List[RecognizerResult]:
+        """Filtre les résultats en supprimant les termes de la allow-list"""
+        if not allow_list_terms:
+            return results
+        
+        filtered_results = []
+        allow_list_lower = [term.lower().strip() for term in allow_list_terms]
+        
+        for result in results:
+            entity_text = text[result.start:result.end].lower().strip()
+            
+            # Garder l'entité si elle n'est pas dans la allow-list
+            if entity_text not in allow_list_lower:
+                filtered_results.append(result)
+            else:
+                logger.debug(f"🚫 Entité filtrée (allow-list): '{entity_text}'")
+        
+        logger.info(f"🔍 Filtrage allow-list: {len(results)} -> {len(filtered_results)} entités")
+        return filtered_results
\ No newline at end of file
diff --git a/post_processors/__init__.py b/post_processors/__init__.py
new file mode 100644
index 0000000..fe32dac
--- /dev/null
+++ b/post_processors/__init__.py
@@ -0,0 +1,4 @@
+from .deduplication_processor import DeduplicationProcessor
+from .overlap_resolver import OverlapResolver
+
+__all__ = ['DeduplicationProcessor', 'OverlapResolver']
\ No newline at end of file
diff --git a/post_processors/deduplication_processor.py b/post_processors/deduplication_processor.py
new file mode 100644
index 0000000..761c431
--- /dev/null
+++ b/post_processors/deduplication_processor.py
@@ -0,0 +1,66 @@
+from typing import List
+from presidio_analyzer import RecognizerResult
+import logging
+
+logger = logging.getLogger(__name__)
+
+class DeduplicationProcessor:
+    def __init__(self):
+        self.rules = [
+            LocationAddressRule()
+        ]
+        logger.info("🔧 DeduplicationProcessor initialisé avec les règles de déduplication")
+    
+    def process(self, results: List[RecognizerResult], text: str) -> List[RecognizerResult]:
+        """Applique les règles de déduplication aux résultats"""
+        processed_results = results.copy()
+        
+        for rule in self.rules:
+            processed_results = rule.apply(processed_results, text)
+            
+        logger.info(f"🔧 DeduplicationProcessor: {len(results)} -> {len(processed_results)} entités")
+        return processed_results
+
+class LocationAddressRule:
+    """Règle pour éviter les doublons entre LOCATION et ADDRESS"""
+    
+    def __init__(self):
+        self.insignificant_terms = {'le', 'la', 'les', 'de', 'du', 'des', 'à', 'au', 'aux'}
+    
+    def apply(self, results: List[RecognizerResult], text: str) -> List[RecognizerResult]:
+        """Supprime les LOCATION qui sont des doublons d'ADDRESS"""
+        locations = [r for r in results if r.entity_type == 'LOCATION']
+        addresses = [r for r in results if r.entity_type == 'ADDRESS']
+        others = [r for r in results if r.entity_type not in ['LOCATION', 'ADDRESS']]
+        
+        filtered_locations = []
+        for location in locations:
+            if self._should_keep_location(location, addresses, text):
+                filtered_locations.append(location)
+            else:
+                location_text = text[location.start:location.end]
+                logger.debug(f"🗑️ Suppression LOCATION dupliquée: '{location_text}'")
+        
+        return addresses + filtered_locations + others
+    
+    def _should_keep_location(self, location: RecognizerResult, addresses: List[RecognizerResult], text: str) -> bool:
+        location_text = text[location.start:location.end].strip().lower()
+        
+        # Ignorer termes non significatifs
+        if (len(location_text) <= 3 or 
+            location_text in self.insignificant_terms):
+            return False
+        
+        # Vérifier chevauchement avec adresses
+        for address in addresses:
+            if self._is_overlapping_or_contained(location, address, text):
+                return False
+        
+        return True
+    
+    def _is_overlapping_or_contained(self, loc: RecognizerResult, addr: RecognizerResult, text: str) -> bool:
+        """Vérifie si une location est contenue dans une address"""
+        loc_text = text[loc.start:loc.end].strip().lower()
+        addr_text = text[addr.start:addr.end].strip().lower()
+        
+        return loc_text in addr_text
\ No newline at end of file
diff --git a/post_processors/overlap_resolver.py b/post_processors/overlap_resolver.py
new file mode 100644
index 0000000..350faed
--- /dev/null
+++ b/post_processors/overlap_resolver.py
@@ -0,0 +1,241 @@
+from typing import List
+from presidio_analyzer import RecognizerResult
+import logging
+import re
+
+logger = logging.getLogger(__name__)
+
+class OverlapResolver:
+    """
+    Résout les chevauchements entre entités de différents types
+    Priorités: IBAN > EMAIL > PHONE > IP_ADDRESS > ADDRESS > LOCATION > ORGANIZATION > PERSON
+    """
+    
+    def __init__(self):
+        # Ordre de priorité (plus haut = plus prioritaire)
+        self.priority_order = {
+            'IBAN': 100,
+            'CREDIT_CARD': 95,
+            'EMAIL_ADDRESS': 90,
+            'BE_ENTERPRISE_NUMBER': 88,  
+            'PHONE_NUMBER': 85,
+            'BE_PHONE_NUMBER': 85,
+            'IP_ADDRESS': 82,
+            'BE_ADDRESS': 75,
+            'FR_ADDRESS': 75,
+            'DATE_TIME': 70,
+            'ORGANIZATION': 65,
+            'LOCATION': 60,
+            'PERSON': 50,
+            'NRP': 40,
+            'URL': 35
+        }
+        
+        # Patterns pour identifier les organisations
+        self.organization_patterns = [
+            r'\\b\\w+Consult\\b',
+            r'\\bSPRL\\s+\\w+\\b',  # Pattern pour SPRL + nom
+            r'\\bSRL\\s+\\w+\\b',   # Pattern pour SRL + nom
+            r'\\bSA\\s+\\w+\\b',    # Pattern pour SA + nom
+            r'\\bASBL\\s+\\w+\\b',  # Pattern pour ASBL + nom
+            r'\\bSCS\\s+\\w+\\b',   # Pattern pour SCS + nom
+            r'\\bSNC\\s+\\w+\\b',   # Pattern pour SNC + nom
+            r'\\bSPRL\\b',
+            r'\\bSRL\\b',
+            r'\\bSA\\b',
+            r'\\bASBL\\b',
+            r'\\bSCS\\b',
+            r'\\bSNC\\b',
+            r'\\bLtd\\b',
+            r'\\bInc\\b',
+            r'\\bCorp\\b',
+            r'\\bGmbH\\b'
+        ]
+        
+        logger.info(f"✅ OverlapResolver initialisé avec {len(self.priority_order)} types d'entités")
+    
+    def process(self, results: List[RecognizerResult], text: str = "") -> List[RecognizerResult]:
+        """
+        Résout les chevauchements en gardant l'entité la plus prioritaire
+        """
+        if not results:
+            return results
+        
+        original_count = len(results)
+        
+        # Appliquer les corrections spécifiques avant résolution des chevauchements
+        corrected_results = self._apply_specific_corrections(results, text)
+        
+        # Trier par position pour traitement séquentiel
+        sorted_results = sorted(corrected_results, key=lambda x: (x.start, x.end))
+        
+        resolved_results = []
+        i = 0
+        
+        while i < len(sorted_results):
+            current = sorted_results[i]
+            overlapping_group = [current]
+            
+            # Trouver tous les chevauchements avec l'entité courante
+            j = i + 1
+            while j < len(sorted_results):
+                if self._is_overlapping(current, sorted_results[j]):
+                    overlapping_group.append(sorted_results[j])
+                    j += 1
+                elif sorted_results[j].start >= current.end:
+                    # Plus de chevauchement possible
+                    break
+                else:
+                    j += 1
+            
+            # Résoudre le groupe de chevauchements
+            if len(overlapping_group) > 1:
+                winner = self._resolve_overlap_group(overlapping_group, text)
+                resolved_results.append(winner)
+                # Avancer l'index pour éviter de retraiter les entités du groupe
+                i = j
+            else:
+                resolved_results.append(current)
+                i += 1
+        
+        logger.info(f"🔧 OverlapResolver: {original_count} -> {len(resolved_results)} entités")
+        return resolved_results
+    
+    def _apply_specific_corrections(self, results: List[RecognizerResult], text: str) -> List[RecognizerResult]:
+        """
+        Applique des corrections spécifiques avant la résolution des chevauchements
+        """
+        corrected_results = []
+        
+        for result in results:
+            entity_text = text[result.start:result.end] if text else ""
+            
+            # Correction 1: PERSON -> ORGANIZATION pour les noms d'entreprise
+            if result.entity_type == 'PERSON' and self._is_organization_name(entity_text):
+                corrected_result = RecognizerResult(
+                    entity_type='ORGANIZATION',
+                    start=result.start,
+                    end=result.end,
+                    score=result.score + 0.1  # Bonus de confiance
+                )
+                logger.debug(f"🔄 Correction PERSON -> ORGANIZATION: '{entity_text}'")
+                corrected_results.append(corrected_result)
+            
+            # Correction 2: Séparer IP des adresses physiques
+            elif result.entity_type in ['BE_ADDRESS', 'FR_ADDRESS'] and self._contains_ip_address(entity_text):
+                # Extraire l'IP et créer une entité séparée
+                ip_matches = list(re.finditer(r'\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b', entity_text))
+                if ip_matches:
+                    for ip_match in ip_matches:
+                        ip_start = result.start + ip_match.start()
+                        ip_end = result.start + ip_match.end()
+                        
+                        # Créer l'entité IP
+                        ip_result = RecognizerResult(
+                            entity_type='IP_ADDRESS',
+                            start=ip_start,
+                            end=ip_end,
+                            score=0.95
+                        )
+                        corrected_results.append(ip_result)
+                        logger.debug(f"🔄 IP extraite de l'adresse: '{ip_match.group()}'")
+                    
+                    # Créer une nouvelle entité adresse SANS la partie IP
+                    # Chercher la partie adresse physique (après l'IP)
+                    address_pattern = r'\b(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève|Clos|Allée)\b.*?\b[1-9]\d{3}\s+[A-Za-zà-ÿ\'-]+'
+                    address_match = re.search(address_pattern, entity_text, re.IGNORECASE)
+                    
+                    if address_match:
+                        address_start = result.start + address_match.start()
+                        address_end = result.start + address_match.end()
+                        
+                        # Vérifier qu'il n'y a pas de chevauchement avec l'IP
+                        ip_overlaps = any(not (address_end <= ip_start or address_start >= ip_end) 
+                                        for ip_match in ip_matches 
+                                        for ip_start, ip_end in [(result.start + ip_match.start(), result.start + ip_match.end())])
+                        
+                        if not ip_overlaps:
+                            address_result = RecognizerResult(
+                                entity_type=result.entity_type,
+                                start=address_start,
+                                end=address_end,
+                                score=result.score
+                            )
+                            corrected_results.append(address_result)
+                            logger.debug(f"🔄 Adresse physique séparée: '{address_match.group()}'")
+                else:
+                    corrected_results.append(result)
+            else:
+                corrected_results.append(result)
+        
+        return corrected_results
+    
+    def _is_organization_name(self, text: str) -> bool:
+        """
+        Détermine si un texte ressemble à un nom d'organisation
+        """
+        for pattern in self.organization_patterns:
+            if re.search(pattern, text, re.IGNORECASE):
+                return True
+        return False
+    
+    def _contains_ip_address(self, text: str) -> bool:
+        """
+        Vérifie si le texte contient une adresse IP
+        """
+        ip_pattern = r'\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b'
+        return bool(re.search(ip_pattern, text))
+    
+    def _is_overlapping(self, entity1: RecognizerResult, entity2: RecognizerResult) -> bool:
+        """
+        Vérifie si deux entités se chevauchent
+        """
+        return not (entity1.end <= entity2.start or entity1.start >= entity2.end)
+    
+    def _resolve_overlap_group(self, overlapping_entities: List[RecognizerResult], text: str = "") -> RecognizerResult:
+        """
+        Résout un groupe d'entités qui se chevauchent
+        Critères: 1) Priorité du type, 2) Score de confiance, 3) Longueur
+        """
+        def get_priority_score(entity):
+            base_priority = self.priority_order.get(entity.entity_type, 0)
+            confidence_bonus = entity.score * 10  # Score 0.9 = +9 points
+            
+            # Calculer la longueur depuis les positions
+            entity_length = entity.end - entity.start
+            length_bonus = entity_length * 0.1  # Bonus longueur
+            
+            # Bonus spécial pour IBAN vs FR_DRIVER_LICENSE
+            if entity.entity_type == 'IBAN':
+                # Vérifier si c'est un vrai IBAN (commence par code pays)
+                if text:
+                    entity_text = text[entity.start:entity.end].replace(' ', '')
+                    if re.match(r'^[A-Z]{2}[0-9]{2}', entity_text):
+                        base_priority += 20  # Bonus pour vrai IBAN
+            
+            return base_priority + confidence_bonus + length_bonus
+        
+        # Trier par score de priorité décroissant
+        sorted_entities = sorted(overlapping_entities, 
+                               key=get_priority_score, 
+                               reverse=True)
+        
+        winner = sorted_entities[0]
+        
+        # Log des entités écartées (si texte disponible)
+        if text:
+            for loser in sorted_entities[1:]:
+                loser_text = text[loser.start:loser.end]
+                logger.debug(f"❌ Écarté: {loser.entity_type} '{loser_text}' (score: {get_priority_score(loser):.1f})")
+            
+            winner_text = text[winner.start:winner.end]
+            logger.debug(f"✅ Gagnant: {winner.entity_type} '{winner_text}' (score: {get_priority_score(winner):.1f})")
+        
+        return winner
+    
+    def add_entity_priority(self, entity_type: str, priority: int):
+        """
+        Ajoute ou modifie la priorité d'un type d'entité
+        """
+        self.priority_order[entity_type] = priority
+        logger.info(f"📊 Priorité mise à jour: {entity_type} = {priority}")
\ No newline at end of file
diff --git a/refiners/__init__.py b/refiners/__init__.py
new file mode 100644
index 0000000..2bcc8fb
--- /dev/null
+++ b/refiners/__init__.py
@@ -0,0 +1 @@
+# Refiners package
\ No newline at end of file
diff --git a/refiners/date_refiner.py b/refiners/date_refiner.py
new file mode 100644
index 0000000..bc0c87a
--- /dev/null
+++ b/refiners/date_refiner.py
@@ -0,0 +1,89 @@
+from abc import ABC, abstractmethod
+from typing import Optional, Tuple
+import re
+import logging
+
+logger = logging.getLogger(__name__)
+
+class EntityRefiner(ABC):
+    """Classe de base pour le recadrage d'entités"""
+    
+    def __init__(self, entity_type: str):
+        self.entity_type = entity_type
+    
+    @abstractmethod
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Recadre une entité détectée"""
+        pass
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Vérifie si ce raffineur doit traiter ce type d'entité"""
+        return entity_type == self.entity_type
+
+class DateRefiner(EntityRefiner):
+    """Raffineur pour les dates - élimine les faux positifs"""
+    
+    def __init__(self):
+        super().__init__("DATE_TIME")
+        # Patterns pour valider les vraies dates
+        self.valid_date_patterns = [
+            # Format DD/MM/YYYY
+            re.compile(r"\b(?:0[1-9]|[12][0-9]|3[01])/(?:0[1-9]|1[0-2])/(?:19|20)\d{2}\b"),
+            # Format DD-MM-YYYY
+            re.compile(r"\b(?:0[1-9]|[12][0-9]|3[01])-(?:0[1-9]|1[0-2])-(?:19|20)\d{2}\b"),
+            # Format ISO YYYY-MM-DD
+            re.compile(r"\b(?:19|20)\d{2}-(?:0[1-9]|1[0-2])-(?:0[1-9]|[12][0-9]|3[01])\b"),
+            # Dates avec mois en lettres
+            re.compile(r"\b(?:0?[1-9]|[12][0-9]|3[01])\s+(?:janvier|février|mars|avril|mai|juin|juillet|août|septembre|octobre|novembre|décembre)\s+(?:19|20)\d{2}\b", re.IGNORECASE),
+            # Heures
+            re.compile(r"\b(?:[01][0-9]|2[0-3]):[0-5][0-9](?::[0-5][0-9])?\b")
+        ]
+        
+        # Patterns à rejeter (faux positifs courants)
+        self.reject_patterns = [
+            # Codes IBAN belges (BE + chiffres)
+            re.compile(r"\bBE\d{2,}\b", re.IGNORECASE),
+            # Numéros d'entreprise belges
+            re.compile(r"\bBE\d{3}\.\d{3}\.\d{3}\b"),
+            # Mots comme HTVA, TVA, etc.
+            re.compile(r"\b(?:HTVA|TVA|BCE|ONSS|SIREN|SIRET)\b", re.IGNORECASE),
+            # Données sensibles (texte)
+            re.compile(r"\b(?:données?\s+sensibles?)\b", re.IGNORECASE),
+            # Codes postaux isolés
+            re.compile(r"^\d{4}$"),
+            # Codes courts (2-4 caractères alphanumériques)
+            re.compile(r"^[A-Z]{2}\d{1,2}$")
+        ]
+    
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Valide si l'entité détectée est vraiment une date"""
+        ent_text = text[start:end].strip()
+        
+        # Vérifier si c'est un pattern à rejeter
+        for reject_pattern in self.reject_patterns:
+            if reject_pattern.search(ent_text):
+                logger.info(f"Date rejetée (faux positif): '{ent_text}'")
+                return None
+        
+        # Vérifier si c'est un pattern de date valide
+        for valid_pattern in self.valid_date_patterns:
+            if valid_pattern.search(ent_text):
+                logger.info(f"Date validée: '{ent_text}'")
+                return (start, end)
+        
+        # Si aucun pattern valide trouvé, rejeter
+        logger.info(f"Date rejetée (format invalide): '{ent_text}'")
+        return None
+    
+    def validate_date_logic(self, day: int, month: int, year: int) -> bool:
+        """Valide la logique de la date (jours/mois corrects)"""
+        if month < 1 or month > 12:
+            return False
+        
+        days_in_month = [31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]
+        
+        # Année bissextile
+        if year % 4 == 0 and (year % 100 != 0 or year % 400 == 0):
+            days_in_month[1] = 29
+        
+        return 1 <= day <= days_in_month[month - 1]
\ No newline at end of file
diff --git a/refiners/iban_refiner.py b/refiners/iban_refiner.py
new file mode 100644
index 0000000..87b8290
--- /dev/null
+++ b/refiners/iban_refiner.py
@@ -0,0 +1,49 @@
+from abc import ABC, abstractmethod
+from typing import Optional, Tuple
+import re
+import logging
+
+logger = logging.getLogger(__name__)
+
+class EntityRefiner(ABC):
+    """Classe de base pour le recadrage d'entités"""
+    
+    def __init__(self, entity_type: str):
+        self.entity_type = entity_type
+    
+    @abstractmethod
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Recadre une entité détectée"""
+        pass
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Vérifie si ce raffineur doit traiter ce type d'entité"""
+        return entity_type == self.entity_type
+
+class IBANRefiner(EntityRefiner):
+    """Raffineur pour les IBAN"""
+    
+    def __init__(self):
+        super().__init__("IBAN")
+        self.iban_regex = re.compile(r"\b[A-Z]{2}[0-9]{2}(?:\s[0-9]{4}){3}\b", re.IGNORECASE)
+    
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        ent_text = text[start:end].strip()
+        match = self.iban_regex.search(ent_text)
+        
+        if not match:
+            logger.warning(f"Invalid IBAN detected, skipping: '{ent_text}'")
+            return None
+            
+        true_iban = match.group(0)
+        start_offset = ent_text.find(true_iban)
+        
+        if start_offset == -1:
+            logger.warning(f"IBAN regex match but cannot find substring position: '{ent_text}'")
+            return None
+            
+        new_start = start + start_offset
+        new_end = new_start + len(true_iban)
+        
+        logger.debug(f"Adjusted IBAN span: {start}-{end} => {new_start}-{new_end}")
+        return (new_start, new_end)
\ No newline at end of file
diff --git a/refiners/ip_refiner.py b/refiners/ip_refiner.py
new file mode 100644
index 0000000..650c2d7
--- /dev/null
+++ b/refiners/ip_refiner.py
@@ -0,0 +1,52 @@
+from abc import ABC, abstractmethod
+from typing import Optional, Tuple
+import re
+import logging
+
+logger = logging.getLogger(__name__)
+
+class EntityRefiner(ABC):
+    """Classe de base pour le recadrage d'entités"""
+    
+    def __init__(self, entity_type: str):
+        self.entity_type = entity_type
+    
+    @abstractmethod
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Recadre une entité détectée"""
+        pass
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Vérifie si ce raffineur doit traiter ce type d'entité"""
+        return entity_type == self.entity_type
+
+class IPAddressRefiner(EntityRefiner):
+    """Raffineur pour les adresses IP"""
+    
+    def __init__(self):
+        super().__init__("IP_ADDRESS")
+        self.ipv4_regex = re.compile(
+            r"\b(?:(?:25[0-5]|2[0-4][0-9]|1\d{2}|[1-9]?\d)\.){3}"
+            r"(?:25[0-5]|2[0-4][0-9]|1\d{2}|[1-9]?\d)\b"
+        )
+    
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        ent_text = text[start:end].strip()
+        match = self.ipv4_regex.search(ent_text)
+        
+        if not match:
+            logger.warning(f"Invalid IP detected, skipping: '{ent_text}'")
+            return None
+            
+        true_ip = match.group(0)
+        start_offset = ent_text.find(true_ip)
+        
+        if start_offset == -1:
+            logger.warning(f"IP regex match but cannot find substring position: '{ent_text}'")
+            return None
+            
+        new_start = start + start_offset
+        new_end = new_start + len(true_ip)
+        
+        logger.debug(f"Adjusted IP span: {start}-{end} => {new_start}-{new_end}")
+        return (new_start, new_end)
\ No newline at end of file
diff --git a/refiners/location_address_refiner.py b/refiners/location_address_refiner.py
new file mode 100644
index 0000000..5be60d7
--- /dev/null
+++ b/refiners/location_address_refiner.py
@@ -0,0 +1,76 @@
+from typing import List, Optional, Tuple
+from presidio_analyzer import RecognizerResult
+from abc import ABC, abstractmethod
+import logging
+
+logger = logging.getLogger(__name__)
+
+class EntityRefiner(ABC):
+    """Classe de base pour le recadrage d'entités"""
+    
+    def __init__(self, entity_type: str):
+        self.entity_type = entity_type
+    
+    @abstractmethod
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """
+        Recadre une entité détectée
+        
+        Args:
+            text: Le texte complet
+            start: Position de début de l'entité détectée
+            end: Position de fin de l'entité détectée
+            
+        Returns:
+            Tuple (nouveau_start, nouveau_end) ou None si l'entité doit être ignorée
+        """
+        pass
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Vérifie si ce raffineur doit traiter ce type d'entité"""
+        return entity_type == self.entity_type
+
+class LocationAddressRefiner(EntityRefiner):
+    """
+    Refiner pour filtrer les doublons entre LOCATION et BE_ADDRESS/FR_ADDRESS.
+    Ce refiner ne modifie pas les positions mais peut supprimer des entités.
+    """
+    
+    def __init__(self):
+        super().__init__("LOCATION")  # Ne traite que les LOCATION
+        self.address_entities = {'BE_ADDRESS', 'FR_ADDRESS'}
+        self.location_entity = 'LOCATION'
+        # Cache pour stocker les adresses détectées
+        self._detected_addresses = []
+    
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """
+        Vérifie si cette LOCATION fait partie d'une adresse déjà détectée.
+        
+        Args:
+            text: Le texte complet
+            start: Position de début de la LOCATION
+            end: Position de fin de la LOCATION
+            
+        Returns:
+            Tuple (start, end) si la location doit être conservée, None sinon
+        """
+        location_text = text[start:end].strip().lower()
+        
+        # Ignorer les locations trop courtes ou non significatives
+        if len(location_text) <= 3 or location_text in ['tel', 'fax', 'gsm']:
+            logger.debug(f"Ignoring short/insignificant location: '{location_text}'")
+            return None
+        
+        # Chercher des adresses dans le texte (simple heuristique)
+        # Cette approche est limitée car on n'a accès qu'à une entité à la fois
+        # Une meilleure approche serait de modifier l'architecture globale
+        
+        # Pour l'instant, on garde toutes les locations valides
+        # et on laisse un post-processing global gérer les doublons
+        logger.debug(f"Keeping location: '{location_text}'")
+        return (start, end)
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Ne traite que les entités LOCATION"""
+        return entity_type == self.location_entity
\ No newline at end of file
diff --git a/refiners/word_boundary_refiner.py b/refiners/word_boundary_refiner.py
new file mode 100644
index 0000000..22b6f11
--- /dev/null
+++ b/refiners/word_boundary_refiner.py
@@ -0,0 +1,39 @@
+import re
+from typing import Optional, Tuple
+import logging
+
+logger = logging.getLogger(__name__)
+
+class WordBoundaryRefiner:
+    """Refiner pour étendre les entités aux limites de mots complets"""
+    
+    def __init__(self):
+        self.entity_type = "ALL"  # S'applique à tous les types d'entités
+    
+    def should_process(self, entity_type: str) -> bool:
+        """Ce refiner s'applique à tous les types d'entités"""
+        return True
+    
+    def refine(self, text: str, start: int, end: int) -> Optional[Tuple[int, int]]:
+        """Étend l'entité pour inclure le mot complet"""
+        try:
+            # Trouver le début du mot
+            new_start = start
+            while new_start > 0 and text[new_start - 1].isalnum():
+                new_start -= 1
+            
+            # Trouver la fin du mot
+            new_end = end
+            while new_end < len(text) and text[new_end].isalnum():
+                new_end += 1
+            
+            # Retourner les nouvelles positions si elles ont changé
+            if new_start != start or new_end != end:
+                logger.debug(f"Extended entity boundaries from [{start}:{end}] to [{new_start}:{new_end}]")
+                return (new_start, new_end)
+            
+            return None
+            
+        except Exception as e:
+            logger.error(f"Error in WordBoundaryRefiner: {e}")
+            return None
\ No newline at end of file