{ "source_file": "CRH 23097531.pdf", "total_replacements": 194, "regex_replacements": 119, "ner_replacements": 18, "sweep_replacements": 57, "entities_found": [ { "original": "10003854436", "replacement": "[CODE_BARRE_1]", "source": "regex", "category": "code_barre" }, { "original": "secr.pneumo@ch-cotebasque.fr", "replacement": "[EMAIL_1]", "source": "regex", "category": "email" }, { "original": "sschneider@ch-cotebasque.fr", "replacement": "[EMAIL_2]", "source": "regex", "category": "email" }, { "original": "msabatini@ch-cotebasque.fr", "replacement": "[EMAIL_3]", "source": "regex", "category": "email" }, { "original": "prigaud@ch-cotebasque.fr", "replacement": "[EMAIL_4]", "source": "regex", "category": "email" }, { "original": "cnocent@ch-cotebasque.fr", "replacement": "[EMAIL_5]", "source": "regex", "category": "email" }, { "original": "jpmathieu@ch-cotebasque.fr", "replacement": "[EMAIL_6]", "source": "regex", "category": "email" }, { "original": "lmasse@ch-cotebasque.fr", "replacement": "[EMAIL_7]", "source": "regex", "category": "email" }, { "original": "clethrosne@ch-cotebasque.fr", "replacement": "[EMAIL_8]", "source": "regex", "category": "email" }, { "original": "dbonnet@ch-cotebasque.fr", "replacement": "[EMAIL_9]", "source": "regex", "category": "email" }, { "original": "eabraham@ch-cotebasque.fr", "replacement": "[EMAIL_10]", "source": "regex", "category": "email" }, { "original": "eellie@ch-cotebasque.fr", "replacement": "[EMAIL_11]", "source": "regex", "category": "email" }, { "original": "secr.pneumo@ch-cotebasque.fr", "replacement": "[EMAIL_1]", "source": "regex", "category": "email" }, { "original": "sschneider@ch-cotebasque.fr", "replacement": "[EMAIL_2]", "source": "regex", "category": "email" }, { "original": "msabatini@ch-cotebasque.fr", "replacement": "[EMAIL_3]", "source": "regex", "category": "email" }, { "original": "prigaud@ch-cotebasque.fr", "replacement": "[EMAIL_4]", "source": "regex", "category": "email" }, { "original": "cnocent@ch-cotebasque.fr", "replacement": "[EMAIL_5]", "source": "regex", "category": "email" }, { "original": "jpmathieu@ch-cotebasque.fr", "replacement": "[EMAIL_6]", "source": "regex", "category": "email" }, { "original": "lmasse@ch-cotebasque.fr", "replacement": "[EMAIL_7]", "source": "regex", "category": "email" }, { "original": "clethrosne@ch-cotebasque.fr", "replacement": "[EMAIL_8]", "source": "regex", "category": "email" }, { "original": "dbonnet@ch-cotebasque.fr", "replacement": "[EMAIL_9]", "source": "regex", "category": "email" }, { "original": "eabraham@ch-cotebasque.fr", "replacement": "[EMAIL_10]", "source": "regex", "category": "email" }, { "original": "eellie@ch-cotebasque.fr", "replacement": "[EMAIL_11]", "source": "regex", "category": "email" }, { "original": "secr.pneumo@ch-cotebasque.fr", "replacement": "[EMAIL_1]", "source": "regex", "category": "email" }, { "original": "sschneider@ch-cotebasque.fr", "replacement": "[EMAIL_2]", "source": "regex", "category": "email" }, { "original": "msabatini@ch-cotebasque.fr", "replacement": "[EMAIL_3]", "source": "regex", "category": "email" }, { "original": "prigaud@ch-cotebasque.fr", "replacement": "[EMAIL_4]", "source": "regex", "category": "email" }, { "original": "cnocent@ch-cotebasque.fr", "replacement": "[EMAIL_5]", "source": "regex", "category": "email" }, { "original": "jpmathieu@ch-cotebasque.fr", "replacement": "[EMAIL_6]", "source": "regex", "category": "email" }, { "original": "lmasse@ch-cotebasque.fr", "replacement": "[EMAIL_7]", "source": "regex", "category": "email" }, { "original": "clethrosne@ch-cotebasque.fr", "replacement": "[EMAIL_8]", "source": "regex", "category": "email" }, { "original": "dbonnet@ch-cotebasque.fr", "replacement": "[EMAIL_9]", "source": "regex", "category": "email" }, { "original": "eabraham@ch-cotebasque.fr", "replacement": "[EMAIL_10]", "source": "regex", "category": "email" }, { "original": "eellie@ch-cotebasque.fr", "replacement": "[EMAIL_11]", "source": "regex", "category": "email" }, { "original": "[PATIENT_", "replacement": "[PERSONNE_1]", "source": "ner", "score": 0.9068517684936523 }, { "original": "[MEDECIN_2", "replacement": "[PERSONNE_2]", "source": "ner", "score": 0.917240560054779 }, { "original": "[PATIENT", "replacement": "[PERSONNE_3]", "source": "ner", "score": 0.8825443983078003 }, { "original": "[MEDECIN_16", "replacement": "[PERSONNE_4]", "source": "ner", "score": 0.9489185214042664 }, { "original": "[MEDECIN_28", "replacement": "[PERSONNE_5]", "source": "ner", "score": 0.9357218146324158 }, { "original": "Imeth", "replacement": "[PERSONNE_6]", "source": "ner", "score": 0.9053090214729309 }, { "original": "[MEDECIN_31", "replacement": "[PERSONNE_7]", "source": "ner", "score": 0.9426769614219666 }, { "original": "[MEDECIN_32", "replacement": "[PERSONNE_8]", "source": "ner", "score": 0.9661399722099304 }, { "original": "[MEDECIN_22", "replacement": "[PERSONNE_9]", "source": "ner", "score": 0.9413198828697205 }, { "original": "[MEDECIN_33", "replacement": "[PERSONNE_10]", "source": "ner", "score": 0.9509629011154175 }, { "original": "[MEDECIN_34", "replacement": "[PERSONNE_11]", "source": "ner", "score": 0.9420625567436218 }, { "original": "Pierre Rigaud", "replacement": "[MEDECIN_11]", "source": "ner", "score": 0.9939513206481934 }, { "original": "J. LOEB", "replacement": "[PERSONNE_12]", "source": "ner", "score": 0.9422516226768494 }, { "original": "[MEDECIN", "replacement": "[PERSONNE_13]", "source": "ner", "score": 0.9810889959335327 }, { "original": "[MEDECIN_", "replacement": "[PERSONNE_14]", "source": "ner", "score": 0.9928128123283386 }, { "original": "[PATIENT_3", "replacement": "[PERSONNE_15]", "source": "ner", "score": 0.8995981812477112 }, { "original": "LESCAT Anne-marie", "replacement": "[PERSONNE_16]", "source": "ner", "score": 0.9849374294281006 }, { "original": "[MEDECIN_42", "replacement": "[PERSONNE_17]", "source": "ner", "score": 0.8268469572067261 } ] }