{
"domain": "publishing",
"entities": [
{
"name": "author",
"weight": 5,
"columns": [
{
"name": "author_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "middle_name",
"type": "string",
"null_rate_default": 0.6
},
{
"name": "birth_date",
"type": "date",
"null_rate_default": 0.3
},
{
"name": "language",
"type": "string",
"pool_name": "language"
},
{
"name": "email",
"type": "string",
"null_rate_default": 0.15
},
{
"name": "orcid",
"type": "string",
"null_rate_default": 0.4
},
{
"name": "biography",
"type": "string",
"null_rate_default": 0.5
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "publisher",
"weight": 1,
"columns": [
{
"name": "publisher_id",
"type": "string"
},
{
"name": "publisher_name",
"type": "string",
"pool_name": "company"
},
{
"name": "address_line1",
"type": "string"
},
{
"name": "city",
"type": "string"
},
{
"name": "country",
"type": "string"
},
{
"name": "website",
"type": "string"
},
{
"name": "email",
"type": "string",
"null_rate_default": 0.2
},
{
"name": "phone",
"type": "string",
"null_rate_default": 0.4
},
{
"name": "founding_year",
"type": "int"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "book",
"weight": 20,
"columns": [
{
"name": "book_id",
"type": "string"
},
{
"name": "title",
"type": "string",
"pool_name": "product_names"
},
{
"name": "isbn",
"type": "string",
"null_rate_default": 0.05
},
{
"name": "issn",
"type": "string",
"null_rate_default": 0.85
},
{
"name": "doi",
"type": "string",
"null_rate_default": 0.5
},
{
"name": "edition",
"type": "int"
},
{
"name": "publication_date",
"type": "date"
},
{
"name": "language",
"type": "string",
"pool_name": "language"
},
{
"name": "pages",
"type": "string"
},
{
"name": "volume",
"type": "string",
"null_rate_default": 0.7
},
{
"name": "subject_category",
"type": "string",
"pool_name": "publication_type"
},
{
"name": "author_id",
"type": "string"
},
{
"name": "publisher_id",
"type": "string"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "author_id",
"target_entity": "author"
},
{
"source_col": "publisher_id",
"target_entity": "publisher"
}
]
},
{
"name": "library_catalog",
"weight": 50,
"columns": [
{
"name": "catalog_id",
"type": "string"
},
{
"name": "book_id",
"type": "string"
},
{
"name": "branch_location",
"type": "string",
"pool_name": "city"
},
{
"name": "acquisition_date",
"type": "date"
},
{
"name": "copies_owned",
"type": "int"
},
{
"name": "shelf_location",
"type": "string"
},
{
"name": "status",
"type": "string",
"pool_name": "status_active"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "book_id",
"target_entity": "book"
}
]
},
{
"name": "patron",
"weight": 80,
"columns": [
{
"name": "patron_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "email",
"type": "string"
},
{
"name": "phone",
"type": "string",
"null_rate_default": 0.3
},
{
"name": "membership_date",
"type": "date"
},
{
"name": "membership_tier",
"type": "string",
"pool_name": "tier"
},
{
"name": "status",
"type": "string",
"pool_name": "status_active"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
}
],
"hn_types": [
{
"name": "same_isbn_diff_title",
"entity_type": "book",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"isbn\"], \"attr_fields\": [\"title\", \"edition\", \"publication_date\", \"publisher_id\"]}"
},
{
"name": "same_title_diff_author",
"entity_type": "book",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"title\"], \"attr_fields\": [\"author_id\", \"publisher_id\", \"publication_date\", \"isbn\"]}"
},
{
"name": "same_author_same_subject",
"entity_type": "book",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"author_id\", \"subject_category\"], \"attr_fields\": [\"title\", \"isbn\", \"publication_date\", \"publisher_id\"]}"
},
{
"name": "same_name_diff_patron",
"entity_type": "patron",
"config_json": "{\"pattern\": \"same_name_different_everything\", \"first_name_col\": \"first_name\", \"last_name_col\": \"last_name\", \"attr_fields\": [\"email\", \"phone\", \"membership_date\", \"membership_tier\"]}"
},
{
"name": "same_email_diff_patron",
"entity_type": "patron",
"config_json": "{\"pattern\": \"same_email\", \"email_col\": \"email\", \"attr_fields\": [\"first_name\", \"last_name\", \"phone\", \"membership_tier\"]}"
}
]
}