create-entity-server 0.0.26 → 0.0.31

package/template/templates/ocr/business_reg.json

@@ -0,0 +1,145 @@
+{
+    "doc_type": "business_reg",
+    "label": "사업자등록증",
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 5,
+        "x_tolerance": 10,
+        "remove_patterns": [
+            "NATIONAL TAX SERVICE REPUBLIC OF KOREA",
+            "나라사랑e",
+            "nts\\.go\\.kr"
+        ],
+        "remove_keywords": [
+            "KOREA",
+            "OF",
+            "NATIONAL",
+            "REPULIC",
+            "국세청",
+            "nts.go.kr",
+            "TAX",
+            "SERVICE"
+        ],
+        "merge_patterns": ["주식회사", "(주)"]
+    },
+    "keywords": {
+        "dummy": ["국세청", "법인사업자", "개인사업자", "사업자등록증"],
+        "corp_num": ["등록번호", "사업자등록번호"],
+        "corp_reg_num": ["법인등록번호"],
+        "corp_name": ["법인명", "단체명", "상호", "법인명(단체명)"],
+        "ceo_name": ["대표자", "성명"],
+        "birthday": ["생년월일"],
+        "open_date": ["개업연월일"],
+        "address": ["사업장소재지"],
+        "head_address": ["본점소재지"],
+        "biz_type": ["업태", "사업의종류"],
+        "biz_class": ["종목"],
+        "issue_reason": ["발급사유"],
+        "tax_unit": ["사업자단위과세적용사업자여부"],
+        "email": ["전자세금계산서전용전자우편주소"]
+    },
+    "fields": {
+        "corp_num": {
+            "mode": "regex_scan",
+            "pattern": "([0-9]{3}-[0-9]{2}-[0-9]{5})",
+            "refine": "biz_num"
+        },
+        "corp_reg_num": {
+            "mode": "anchor",
+            "required_keyword": "법인등록번호",
+            "criteria": [
+                { "keyword": "법인등록번호", "direction": "RIGHT" },
+                { "keyword": "대표자", "direction": "DOWN" },
+                { "keyword": "상호", "direction": "DOWN" },
+                { "keyword": "사업장소재지", "direction": "UP" }
+            ]
+        },
+        "corp_name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "법인명(단체명)", "direction": "RIGHT" },
+                { "keyword": "상호", "direction": "RIGHT" },
+                { "keyword": "등록번호", "direction": "DOWN" },
+                { "keyword": "대표자", "direction": "UP" },
+                { "keyword": "성명", "direction": "UP" },
+                { "keyword": "생년월일", "direction": "LEFT" }
+            ]
+        },
+        "ceo_name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "대표자", "direction": "RIGHT" },
+                { "keyword": "성명", "direction": "RIGHT" },
+                { "keyword": "생년월일", "direction": "LEFT" },
+                { "keyword": "법인명(단체명)", "direction": "DOWN" },
+                { "keyword": "상호", "direction": "DOWN" },
+                { "keyword": "개업연월일", "direction": "UP" }
+            ]
+        },
+        "open_date": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "개업연월일", "direction": "RIGHT" },
+                { "keyword": "법인등록번호", "direction": "LEFT" },
+                { "keyword": "사업장소재지", "direction": "UP" },
+                { "keyword": "대표자", "direction": "DOWN" },
+                { "keyword": "성명", "direction": "DOWN" }
+            ],
+            "refine": "date"
+        },
+        "address": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "사업장소재지", "direction": "RIGHT" },
+                { "keyword": "사업장소재지", "direction": "DOWN" },
+                { "keyword": "본점소재지", "direction": "UP" },
+                { "keyword": "사업의종류", "direction": "UP" },
+                { "keyword": "개업연월일", "direction": "DOWN" }
+            ]
+        },
+        "biz_type": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "업태", "direction": "RIGHT" },
+                { "keyword": "종목", "direction": "LEFT" },
+                { "keyword": "본점소재지", "direction": "DOWN" },
+                { "keyword": "사업장소재지", "direction": "DOWN" },
+                { "keyword": "발급사유", "direction": "UP" },
+                { "keyword": "사업자단위과세적용사업자여부", "direction": "UP" }
+            ]
+        },
+        "biz_class": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "종목", "direction": "RIGHT" },
+                { "keyword": "본점소재지", "direction": "DOWN" },
+                { "keyword": "사업장소재지", "direction": "DOWN" },
+                { "keyword": "발급사유", "direction": "UP" },
+                { "keyword": "사업자단위과세적용사업자여부", "direction": "UP" }
+            ]
+        },
+        "email": {
+            "mode": "anchor",
+            "criteria": [
+                {
+                    "keyword": "전자세금계산서전용전자우편주소",
+                    "direction": "RIGHT"
+                },
+                {
+                    "keyword": "전자세금계산서전용전자우편주소",
+                    "direction": "DOWN"
+                }
+            ]
+        }
+    },
+    "required_fields": ["corp_num", "corp_name"],
+    "confidence_weight": {
+        "corp_num": 0.35,
+        "corp_name": 0.3,
+        "ceo_name": 0.1,
+        "open_date": 0.1,
+        "address": 0.1,
+        "biz_type": 0.025,
+        "biz_class": 0.025
+    }
+}

package/template/templates/ocr/career_cert.json

@@ -0,0 +1,93 @@
+{
+    "doc_type": "career_cert",
+    "label": "경력증명서 (재직증명서)",
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 5,
+        "x_tolerance": 10
+    },
+    "keywords": {
+        "name": ["성명", "이름", "위\\s*사람"],
+        "id_number": ["주민등록번호", "생년월일"],
+        "company": ["회사명", "법인명", "상호", "근무처"],
+        "department": ["부서", "소속"],
+        "position": ["직위", "직급", "직책"],
+        "join_date": ["입사일", "입사일자", "근무시작일"],
+        "leave_date": ["퇴사일", "퇴사일자", "근무종료일"],
+        "issue_date": ["발급일", "발행일", "증명서발급일"],
+        "issuer": ["대표이사", "대표자", "발급기관"]
+    },
+    "fields": {
+        "name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "성명", "direction": "RIGHT" },
+                { "keyword": "이름", "direction": "RIGHT" }
+            ],
+            "refine": "trim"
+        },
+        "id_number": {
+            "mode": "regex_scan",
+            "pattern": "([0-9]{6}-[0-9*]{7})",
+            "refine": "id_num"
+        },
+        "company": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "회사명", "direction": "RIGHT" },
+                { "keyword": "법인명", "direction": "RIGHT" },
+                { "keyword": "상호", "direction": "RIGHT" }
+            ]
+        },
+        "department": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "부서", "direction": "RIGHT" },
+                { "keyword": "소속", "direction": "RIGHT" }
+            ]
+        },
+        "position": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "직위", "direction": "RIGHT" },
+                { "keyword": "직급", "direction": "RIGHT" },
+                { "keyword": "직책", "direction": "RIGHT" }
+            ]
+        },
+        "join_date": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "입사일", "direction": "RIGHT" },
+                { "keyword": "입사일자", "direction": "RIGHT" }
+            ],
+            "refine": "date"
+        },
+        "leave_date": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "퇴사일", "direction": "RIGHT" },
+                { "keyword": "퇴사일자", "direction": "RIGHT" }
+            ],
+            "refine": "date"
+        },
+        "issue_date": {
+            "mode": "regex_scan",
+            "pattern": "발급일[\\s:]*([0-9]{4}[.\\-/][0-9]{1,2}[.\\-/][0-9]{1,2})",
+            "refine": "date"
+        },
+        "issuer": {
+            "mode": "regex_scan",
+            "pattern": "대표(?:이사)?\\s*([가-힣]{2,5})(?:\\s*[\\(印\\)\\s]|$)"
+        }
+    },
+    "required_fields": ["name", "company"],
+    "confidence_weight": {
+        "name": 0.2,
+        "company": 0.2,
+        "join_date": 0.2,
+        "leave_date": 0.15,
+        "position": 0.1,
+        "issue_date": 0.1,
+        "issuer": 0.05
+    }
+}

package/template/templates/ocr/driver_license.json

@@ -0,0 +1,89 @@
+{
+    "doc_type": "driver_license",
+    "label": "운전면허증",
+
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 4,
+        "x_tolerance": 8,
+        "remove_keywords": ["(Driver's License)", "Driver's", "License"],
+        "remove_patterns": ["^\\s*$", "^[가-힣]{2,4}\\s*도로교통공단$"]
+    },
+
+    "keywords": {
+        "keyword": ["자동차운전면허증", "적성검사", "기간"],
+        "license_type": [
+            "1종대형",
+            "1종보통",
+            "2종보통",
+            "2종소형",
+            "원동기",
+            "특수(대형견인,구난차)",
+            "특수(소형견인,구난차)",
+            "특수(건설기계)",
+            "특수(농기계)",
+            "특수(비상구난차)",
+            "특수(경찰용)",
+            "특수(소형이동기계)"
+        ]
+    },
+
+    "fields": {
+        "name": {
+            "mode": "anchor",
+            "criteria": [
+                {
+                    "keyword": "(?:[가-힣]+\\s+)?\\d{2}-\\d{2}-\\d{6}-\\d{2}",
+                    "direction": "DOWN_OVERLAP",
+                    "is_regex": true
+                },
+                {
+                    "keyword": "\\d{6}-?(\\d{7}|\\*{7})",
+                    "direction": "UP_OVERLAP",
+                    "is_regex": true
+                }
+            ]
+        },
+        "license_type": {
+            "mode": "keywords_scan",
+            "keywords_key": "license_type"
+        },
+        "license_num": {
+            "mode": "regex_scan",
+            "pattern": "(?:[가-힣]+\\s+)?(\\d{2}-\\d{2}-\\d{6}-\\d{2})",
+            "refine": "driver_license"
+        },
+        "id_number": {
+            "mode": "regex_scan",
+            "pattern": "(\\d{6}-?(\\d{7}|\\*{7}))",
+            "refine": "id_num"
+        },
+        "expiry_date": {
+            "mode": "regex_scan",
+            "pattern": "(\\d{4}[.\\-\\/]\\d{1,2}[.\\-\\/]\\d{1,2})\\s*~\\s*(\\d{4}[.\\-\\/]\\d{1,2}[.\\-\\/]\\d{1,2})",
+            "capture_group": 2,
+            "refine": "date"
+        },
+        "issue_date": {
+            "mode": "dates_scan",
+            "dates_index": 2,
+            "refine": "date"
+        },
+        "issuer": {
+            "mode": "regex_scan",
+            "pattern": "([가-힣]+지방경찰청|[가-힣]+경찰청)",
+            "capture_group": 1
+        }
+    },
+
+    "required_fields": ["name", "license_num"],
+
+    "confidence_weight": {
+        "name": 0.25,
+        "id_number": 0.2,
+        "license_num": 0.3,
+        "license_type": 0.1,
+        "issue_date": 0.1,
+        "expiry_date": 0.05
+    }
+}

package/template/templates/ocr/facility_card.json

@@ -0,0 +1,82 @@
+{
+    "doc_type": "facility_card",
+    "label": "건축물대장",
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 5,
+        "x_tolerance": 12
+    },
+    "keywords": {
+        "building_name": ["건물명", "건축물명", "명칭"],
+        "address": ["소재지", "대지위치", "주소"],
+        "building_use": ["주용도", "용도"],
+        "structure": ["구조"],
+        "area": ["연면적", "건축면적", "면적"],
+        "floor_count": ["층수", "지상", "지하"],
+        "approval_date": ["사용승인일", "사용승인", "준공일"],
+        "owner": ["소유자", "건축주"]
+    },
+    "fields": {
+        "building_name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "건물명", "direction": "RIGHT" },
+                { "keyword": "건축물명", "direction": "RIGHT" },
+                { "keyword": "명칭", "direction": "RIGHT" }
+            ]
+        },
+        "address": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "소재지", "direction": "RIGHT" },
+                { "keyword": "대지위치", "direction": "RIGHT" }
+            ]
+        },
+        "building_use": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "주용도", "direction": "RIGHT" },
+                { "keyword": "용도", "direction": "RIGHT" }
+            ]
+        },
+        "structure": {
+            "mode": "anchor",
+            "criteria": [{ "keyword": "구조", "direction": "RIGHT" }]
+        },
+        "area": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "연면적", "direction": "RIGHT" },
+                { "keyword": "건축면적", "direction": "RIGHT" }
+            ],
+            "refine": "number"
+        },
+        "floor_count": {
+            "mode": "regex_scan",
+            "pattern": "지상\\s*(\\d+)층"
+        },
+        "approval_date": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "사용승인일", "direction": "RIGHT" },
+                { "keyword": "사용승인", "direction": "RIGHT" }
+            ],
+            "refine": "date"
+        },
+        "owner": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "소유자", "direction": "RIGHT" },
+                { "keyword": "건축주", "direction": "RIGHT" }
+            ]
+        }
+    },
+    "required_fields": ["address"],
+    "confidence_weight": {
+        "address": 0.3,
+        "building_use": 0.2,
+        "area": 0.2,
+        "approval_date": 0.15,
+        "owner": 0.15
+    }
+}

package/template/templates/ocr/id_card.json

@@ -0,0 +1,55 @@
+{
+    "doc_type": "id_card",
+    "label": "주민등록증",
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 4,
+        "x_tolerance": 8
+    },
+    "keywords": {
+        "name": ["성명", "이름"],
+        "id_number": ["주민등록번호", "등록번호"],
+        "address": ["주소", "거주지"],
+        "issue_date": ["발급일", "발행일"]
+    },
+    "fields": {
+        "name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "성명", "direction": "RIGHT" },
+                { "keyword": "이름", "direction": "RIGHT" }
+            ]
+        },
+        "id_number": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "주민등록번호", "direction": "RIGHT" },
+                { "keyword": "등록번호", "direction": "RIGHT" }
+            ],
+            "refine": "id_num"
+        },
+        "address": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "주소", "direction": "DOWN" },
+                { "keyword": "주소", "direction": "RIGHT" }
+            ]
+        },
+        "issue_date": {
+            "mode": "regex_scan",
+            "pattern": "발급일[\\s:]*([0-9]{4}[.\\-/][0-9]{1,2}[.\\-/][0-9]{1,2})",
+            "refine": "date"
+        },
+        "issuer": {
+            "mode": "regex_scan",
+            "pattern": "(\\S+구청장|\\S+시장|\\S+군수|행정안전부장관|\\S+동장)"
+        }
+    },
+    "required_fields": ["name", "id_number"],
+    "confidence_weight": {
+        "name": 0.25,
+        "id_number": 0.45,
+        "address": 0.2,
+        "issue_date": 0.1
+    }
+}

package/template/templates/ocr/invoice.json

@@ -0,0 +1,92 @@
+{
+    "doc_type": "invoice",
+    "label": "세금계산서",
+
+    "preprocess": {
+        "max_slope": 0.03,
+        "y_tolerance": 5,
+        "x_tolerance": 10,
+        "remove_patterns": ["^\\s*$"]
+    },
+
+    "keywords": {
+        "invoice_num": ["일련번호", "승인번호"],
+        "issuer_corp_num": ["공급자", "등록번호"],
+        "issuer_name": ["공급자", "상호"],
+        "receiver_corp_num": ["공급받는자", "등록번호"],
+        "receiver_name": ["공급받는자", "상호"],
+        "write_date": ["작성일자", "작성 연월일"],
+        "amount_total": ["공급가액", "합계금액"],
+        "tax_total": ["세액"],
+        "total_amount": ["합계금액", "총합계"]
+    },
+
+    "fields": {
+        "invoice_num": {
+            "output_key": "invoice_num",
+            "mode": "regex_scan",
+            "pattern": "([0-9]{8}-[0-9]{8})",
+            "capture_group": 1
+        },
+        "write_date": {
+            "output_key": "write_date",
+            "criteria": [
+                { "keyword": "작성일자", "direction": "RIGHT" },
+                { "keyword": "작성 연월일", "direction": "RIGHT" }
+            ],
+            "refine": "date"
+        },
+        "amount_total": {
+            "output_key": "amount_total",
+            "criteria": [
+                { "keyword": "공급가액", "direction": "RIGHT" },
+                { "keyword": "공급가액", "direction": "DOWN" }
+            ],
+            "refine": "number"
+        },
+        "tax_total": {
+            "output_key": "tax_total",
+            "criteria": [
+                { "keyword": "세액", "direction": "RIGHT" },
+                { "keyword": "세액", "direction": "DOWN" }
+            ],
+            "refine": "number"
+        },
+        "total_amount": {
+            "output_key": "total_amount",
+            "criteria": [
+                { "keyword": "합계금액", "direction": "RIGHT" },
+                { "keyword": "합계금액", "direction": "DOWN" }
+            ],
+            "refine": "number"
+        },
+        "issuer_name": {
+            "output_key": "issuer_name",
+            "region": {
+                "from": "공급자",
+                "to": "공급받는자",
+                "direction": "DOWN",
+                "x_min_anchor": "상호"
+            }
+        },
+        "receiver_name": {
+            "output_key": "receiver_name",
+            "region": {
+                "from": "공급받는자",
+                "to": "작성일자",
+                "direction": "DOWN",
+                "x_min_anchor": "상호"
+            }
+        }
+    },
+
+    "required_fields": ["total_amount"],
+
+    "confidence_weight": {
+        "invoice_num": 0.15,
+        "write_date": 0.15,
+        "amount_total": 0.25,
+        "tax_total": 0.2,
+        "total_amount": 0.25
+    }
+}

package/template/templates/ocr/namecard.json

@@ -0,0 +1,116 @@
+{
+    "doc_type": "namecard",
+    "label": "명함",
+    "preprocess": {
+        "max_slope": 0.05,
+        "y_tolerance": 5,
+        "x_tolerance": 10
+    },
+    "keywords": {
+        "name": ["성명", "이름"],
+        "company": ["주식회사", "(주)", "㈜", "Inc.", "Corp.", "Co.,", "Ltd."],
+        "department": ["부서", "팀", "사업부", "본부", "센터", "실", "TF"],
+        "title": [
+            "대표",
+            "이사",
+            "부장",
+            "차장",
+            "과장",
+            "팀장",
+            "실장",
+            "본부장",
+            "사원",
+            "대리",
+            "수석",
+            "책임",
+            "선임",
+            "CEO",
+            "CTO",
+            "CFO",
+            "COO",
+            "Director",
+            "Manager"
+        ],
+        "address": ["주소", "address", "addr."],
+        "phone": ["Tel", "TEL", "전화", "Mobile", "HP", "직통", "Fax", "FAX"],
+        "email": ["Email", "E-mail", "이메일", "e-mail"],
+        "website": ["www.", "http", "https"]
+    },
+    "fields": {
+        "name": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "성명", "direction": "RIGHT" },
+                { "keyword": "이름", "direction": "RIGHT" }
+            ]
+        },
+        "company": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "주식회사", "direction": "RIGHT" },
+                { "keyword": "주식회사", "direction": "LEFT" },
+                { "keyword": "(주)", "direction": "RIGHT" },
+                { "keyword": "㈜", "direction": "RIGHT" }
+            ]
+        },
+        "department": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "부서", "direction": "RIGHT" },
+                { "keyword": "팀", "direction": "LEFT" },
+                { "keyword": "사업부", "direction": "LEFT" },
+                { "keyword": "본부", "direction": "LEFT" },
+                { "keyword": "센터", "direction": "LEFT" }
+            ]
+        },
+        "title": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "대표이사", "direction": "LEFT" },
+                { "keyword": "대표이사", "direction": "RIGHT" },
+                { "keyword": "이사", "direction": "LEFT" },
+                { "keyword": "부장", "direction": "LEFT" },
+                { "keyword": "차장", "direction": "LEFT" },
+                { "keyword": "과장", "direction": "LEFT" },
+                { "keyword": "팀장", "direction": "LEFT" },
+                { "keyword": "본부장", "direction": "LEFT" },
+                { "keyword": "실장", "direction": "LEFT" },
+                { "keyword": "수석", "direction": "LEFT" },
+                { "keyword": "Director", "direction": "LEFT" },
+                { "keyword": "Manager", "direction": "LEFT" },
+                { "keyword": "CEO", "direction": "LEFT" }
+            ]
+        },
+        "phone": {
+            "mode": "regex_scan",
+            "pattern": "(?:Tel|TEL|전화|Mobile|HP|직통)[\\s.:]*([0-9][0-9\\-\\s]+[0-9])",
+            "multi": true,
+            "refine": "phone"
+        },
+        "email": {
+            "mode": "regex_scan",
+            "pattern": "[a-zA-Z0-9._%+\\-]+@[a-zA-Z0-9.\\-]+\\.[a-zA-Z]{2,}",
+            "multi": true
+        },
+        "address": {
+            "mode": "anchor",
+            "criteria": [
+                { "keyword": "주소", "direction": "RIGHT" },
+                { "keyword": "주소", "direction": "DOWN" },
+                { "keyword": "address", "direction": "RIGHT" }
+            ]
+        },
+        "website": {
+            "mode": "regex_scan",
+            "pattern": "(?:https?://|www\\.)[a-zA-Z0-9.\\-/]+"
+        }
+    },
+    "required_fields": [],
+    "confidence_weight": {
+        "name": 0.3,
+        "company": 0.3,
+        "phone": 0.2,
+        "email": 0.1,
+        "title": 0.1
+    }
+}