{
"domain": "education",
"entities": [
{
"name": "student",
"weight": 100,
"columns": [
{
"name": "student_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "email",
"type": "string"
},
{
"name": "national_student_id",
"type": "string",
"null_rate_default": 0.3
},
{
"name": "previous_institution",
"type": "string",
"null_rate_default": 0.7
},
{
"name": "personal_email",
"type": "string",
"null_rate_default": 0.3
},
{
"name": "phone",
"type": "string",
"null_rate_default": 0.2
},
{
"name": "date_of_birth",
"type": "date"
},
{
"name": "address_line1",
"type": "string"
},
{
"name": "city",
"type": "string"
},
{
"name": "state_code",
"type": "string"
},
{
"name": "postal_code",
"type": "string"
},
{
"name": "enrollment_status",
"type": "string"
},
{
"name": "gpa",
"type": "float",
"null_rate_default": 0.1
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "faculty",
"weight": 2,
"columns": [
{
"name": "faculty_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "email",
"type": "string"
},
{
"name": "department",
"type": "string"
},
{
"name": "title",
"type": "string"
},
{
"name": "office_phone",
"type": "string",
"null_rate_default": 0.3
},
{
"name": "hire_date",
"type": "date"
},
{
"name": "is_tenured",
"type": "boolean"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "course",
"weight": 10,
"columns": [
{
"name": "course_id",
"type": "string"
},
{
"name": "course_name",
"type": "string"
},
{
"name": "faculty_id",
"type": "string"
},
{
"name": "credits",
"type": "int"
},
{
"name": "semester",
"type": "string"
},
{
"name": "max_enrollment",
"type": "int"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "faculty_id",
"target_entity": "faculty"
}
]
},
{
"name": "enrollment",
"weight": 500,
"columns": [
{
"name": "enrollment_id",
"type": "string"
},
{
"name": "student_id",
"type": "string"
},
{
"name": "course_id",
"type": "string"
},
{
"name": "enrollment_date",
"type": "date"
},
{
"name": "grade",
"type": "string"
},
{
"name": "is_dropped",
"type": "boolean"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "student_id",
"target_entity": "student"
},
{
"source_col": "course_id",
"target_entity": "course"
}
]
}
],
"hn_types": [
{
"name": "same_email_diff_student",
"entity_type": "student",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"email\"], \"attr_fields\": [\"first_name\", \"last_name\", \"phone\", \"date_of_birth\"]}"
},
{
"name": "same_name_diff_id",
"entity_type": "student",
"config_json": "{\"pattern\": \"same_name_different_everything\", \"first_name_col\": \"first_name\", \"last_name_col\": \"last_name\", \"attr_fields\": [\"email\", \"date_of_birth\", \"student_id\"]}"
},
{
"name": "twin_same_dob_lastname",
"entity_type": "student",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"date_of_birth\", \"last_name\"], \"attr_fields\": [\"first_name\", \"email\", \"student_id\"]}"
},
{
"name": "student_id_collision_merger",
"entity_type": "student",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"student_id\"], \"attr_fields\": [\"first_name\", \"last_name\", \"date_of_birth\", \"email\"]}"
},
{
"name": "same_faculty_email_diff_name",
"entity_type": "faculty",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"email\"], \"attr_fields\": [\"first_name\", \"last_name\", \"department\"]}"
}
]
}