{ "source_file": "CRH 23056749.pdf", "total_replacements": 122, "regex_replacements": 77, "ner_replacements": 11, "sweep_replacements": 34, "entities_found": [ { "original": "10110601324", "replacement": "[RPPS_1]", "source": "regex", "category": "rpps" }, { "original": "10101096005", "replacement": "[RPPS_2]", "source": "regex", "category": "rpps" }, { "original": "10100163277", "replacement": "[RPPS_3]", "source": "regex", "category": "rpps" }, { "original": "10101639580", "replacement": "[RPPS_4]", "source": "regex", "category": "rpps" }, { "original": "10004606595", "replacement": "[RPPS_5]", "source": "regex", "category": "rpps" }, { "original": "10101480506", "replacement": "[RPPS_6]", "source": "regex", "category": "rpps" }, { "original": "10100858090", "replacement": "[RPPS_7]", "source": "regex", "category": "rpps" }, { "original": "10102268702", "replacement": "[RPPS_8]", "source": "regex", "category": "rpps" }, { "original": "10102272209", "replacement": "[RPPS_9]", "source": "regex", "category": "rpps" }, { "original": "10101856135", "replacement": "[RPPS_10]", "source": "regex", "category": "rpps" }, { "original": "10002815024", "replacement": "[RPPS_11]", "source": "regex", "category": "rpps" }, { "original": "10100817005", "replacement": "[RPPS_12]", "source": "regex", "category": "rpps" }, { "original": "10110601324", "replacement": "[RPPS_1]", "source": "regex", "category": "rpps" }, { "original": "10101096005", "replacement": "[RPPS_2]", "source": "regex", "category": "rpps" }, { "original": "10100163277", "replacement": "[RPPS_3]", "source": "regex", "category": "rpps" }, { "original": "10101639580", "replacement": "[RPPS_4]", "source": "regex", "category": "rpps" }, { "original": "10004606595", "replacement": "[RPPS_5]", "source": "regex", "category": "rpps" }, { "original": "10101480506", "replacement": "[RPPS_6]", "source": "regex", "category": "rpps" }, { "original": "10100858090", "replacement": "[RPPS_7]", "source": "regex", "category": "rpps" }, { "original": "10102268702", "replacement": "[RPPS_8]", "source": "regex", "category": "rpps" }, { "original": "10102272209", "replacement": "[RPPS_9]", "source": "regex", "category": "rpps" }, { "original": "10002815024", "replacement": "[RPPS_11]", "source": "regex", "category": "rpps" }, { "original": "10100817005", "replacement": "[RPPS_12]", "source": "regex", "category": "rpps" }, { "original": "secr.medint@ch-cotebasque.fr", "replacement": "[EMAIL_1]", "source": "regex", "category": "email" }, { "original": "secr.medint@ch-cotebasque.fr", "replacement": "[EMAIL_1]", "source": "regex", "category": "email" }, { "original": "[MEDECIN_", "replacement": "[PERSONNE_1]", "source": "ner", "score": 0.9796693921089172 }, { "original": "Elisabeth GUINARD", "replacement": "[MEDECIN_14]", "source": "ner", "score": 0.9974997043609619 }, { "original": "[MEDEC", "replacement": "[PERSONNE_2]", "source": "ner", "score": 0.9615168571472168 }, { "original": "[MEDECIN_23", "replacement": "[PERSONNE_3]", "source": "ner", "score": 0.9630061984062195 }, { "original": "Kevin DEVE", "replacement": "[MEDECIN_12]", "source": "ner", "score": 0.9974915981292725 }, { "original": "[MEDECIN", "replacement": "[PERSONNE_4]", "source": "ner", "score": 0.9705804586410522 }, { "original": "[MEDECIN_31", "replacement": "[PERSONNE_5]", "source": "ner", "score": 0.9741517305374146 }, { "original": "[MEDECIN_21", "replacement": "[PERSONNE_6]", "source": "ner", "score": 0.96664959192276 }, { "original": "SEGUIN Carmen", "replacement": "[PERSONNE_7]", "source": "ner", "score": 0.9753478765487671 }, { "original": "[MEDECIN_41", "replacement": "[PERSONNE_8]", "source": "ner", "score": 0.8321228623390198 }, { "original": "[MEDECIN_32", "replacement": "[PERSONNE_9]", "source": "ner", "score": 0.8788591027259827 } ] }