{ "source_file": "CRH 23148798.pdf", "total_replacements": 173, "regex_replacements": 114, "ner_replacements": 12, "sweep_replacements": 47, "entities_found": [ { "original": "10100541860", "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_2", "replacement": "[PERSONNE_1]", "source": "ner", "score": 0.9229361414909363 }, { "original": "[MEDECIN", "replacement": "[PERSONNE_2]", "source": "ner", "score": 0.9816927909851074 }, { "original": "[MEDECIN_", "replacement": "[PERSONNE_3]", "source": "ner", "score": 0.99141526222229 }, { "original": "[MEDEC", "replacement": "[PERSONNE_4]", "source": "ner", "score": 0.8633410334587097 }, { "original": "[PATIENT", "replacement": "[PERSONNE_5]", "source": "ner", "score": 0.9630032181739807 }, { "original": "J. LOEB", "replacement": "[PERSONNE_6]", "source": "ner", "score": 0.9365915060043335 }, { "original": "[MEDECIN_12", "replacement": "[PERSONNE_7]", "source": "ner", "score": 0.9372633099555969 }, { "original": "Pierre Rigaud", "replacement": "[MEDECIN_10]", "source": "ner", "score": 0.9751806855201721 }, { "original": "MEDEC", "replacement": "[PERSONNE_8]", "source": "ner", "score": 0.867510974407196 }, { "original": "ne J. LOEB", "replacement": "[PERSONNE_9]", "source": "ner", "score": 0.8669859170913696 }, { "original": "Bricker", "replacement": "[PERSONNE_10]", "source": "ner", "score": 0.9593000411987305 }, { "original": "DE MARI Patrick", "replacement": "[PERSONNE_11]", "source": "ner", "score": 0.9643693566322327 } ] }