{
"domain": "hr",
"entities": [
{
"name": "employee",
"weight": 100,
"columns": [
{
"name": "employee_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "email",
"type": "string"
},
{
"name": "personal_email",
"type": "string",
"null_rate_default": 0.4
},
{
"name": "phone",
"type": "string",
"null_rate_default": 0.15
},
{
"name": "date_of_birth",
"type": "date",
"null_rate_default": 0.1
},
{
"name": "ssn",
"type": "string",
"null_rate_default": 0.03
},
{
"name": "address_line1",
"type": "string"
},
{
"name": "city",
"type": "string"
},
{
"name": "state_code",
"type": "string"
},
{
"name": "postal_code",
"type": "string"
},
{
"name": "job_title",
"type": "string"
},
{
"name": "department_id",
"type": "string"
},
{
"name": "salary",
"type": "float",
"null_rate_default": 0.2
},
{
"name": "hire_date",
"type": "date"
},
{
"name": "employee_type",
"type": "string"
},
{
"name": "work_location",
"type": "string"
},
{
"name": "is_active",
"type": "boolean"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "department_id",
"target_entity": "department"
}
]
},
{
"name": "department",
"weight": 2,
"columns": [
{
"name": "department_id",
"type": "string"
},
{
"name": "department_name",
"type": "string"
},
{
"name": "cost_center",
"type": "string"
},
{
"name": "head_count",
"type": "int"
},
{
"name": "budget",
"type": "float"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "payroll_record",
"weight": 500,
"columns": [
{
"name": "payroll_id",
"type": "string"
},
{
"name": "employee_id",
"type": "string"
},
{
"name": "pay_period_start",
"type": "date"
},
{
"name": "pay_period_end",
"type": "date"
},
{
"name": "gross_pay",
"type": "float"
},
{
"name": "net_pay",
"type": "float"
},
{
"name": "deductions",
"type": "float"
},
{
"name": "hours_worked",
"type": "float"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "employee_id",
"target_entity": "employee"
}
]
},
{
"name": "performance_review",
"weight": 100,
"columns": [
{
"name": "review_id",
"type": "string"
},
{
"name": "employee_id",
"type": "string"
},
{
"name": "review_date",
"type": "date"
},
{
"name": "rating",
"type": "int"
},
{
"name": "reviewer_notes",
"type": "string",
"null_rate_default": 0.4
},
{
"name": "is_promoted",
"type": "boolean"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "employee_id",
"target_entity": "employee"
}
]
}
],
"hn_types": [
{
"name": "same_ssn_diff_employee",
"entity_type": "employee",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"ssn\"], \"attr_fields\": [\"first_name\", \"last_name\", \"email\", \"date_of_birth\"]}"
},
{
"name": "same_email_diff_employee",
"entity_type": "employee",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"email\"], \"attr_fields\": [\"first_name\", \"last_name\", \"employee_id\"]}"
},
{
"name": "same_department_diff_cost_center",
"entity_type": "department",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"department_name\"], \"attr_fields\": [\"cost_center\", \"budget\"]}"
},
{
"name": "same_ssn_name_after_marriage",
"entity_type": "employee",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"ssn\"], \"attr_fields\": [\"first_name\", \"last_name\", \"email\", \"date_of_birth\"]}"
},
{
"name": "same_email_legacy_current",
"entity_type": "employee",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"email\"], \"attr_fields\": [\"first_name\", \"last_name\", \"employee_id\", \"ssn\"]}"
},
{
"name": "same_name_dob_temp_to_perm",
"entity_type": "employee",
"config_json": "{\"pattern\": \"same_name_dob\", \"first_name_col\": \"first_name\", \"last_name_col\": \"last_name\", \"dob_col\": \"date_of_birth\", \"attr_fields\": [\"employee_id\", \"email\", \"ssn\"]}"
}
]
}