{
"domain": "media",
"entities": [
{
"name": "author",
"weight": 30,
"columns": [
{
"name": "author_id",
"type": "string"
},
{
"name": "first_name",
"type": "string"
},
{
"name": "last_name",
"type": "string"
},
{
"name": "orcid",
"type": "string"
},
{
"name": "affiliation",
"type": "string",
"pool_name": "company"
},
{
"name": "email",
"type": "string",
"null_rate_default": 0.15
},
{
"name": "h_index",
"type": "int"
},
{
"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": "issn",
"type": "string"
},
{
"name": "country",
"type": "string",
"pool_name": "country_code"
},
{
"name": "year_founded",
"type": "int"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": []
},
{
"name": "publication",
"weight": 5,
"columns": [
{
"name": "publication_id",
"type": "string"
},
{
"name": "publisher_id",
"type": "string"
},
{
"name": "title",
"type": "string",
"pool_name": "product_names"
},
{
"name": "isbn",
"type": "string"
},
{
"name": "publication_type",
"type": "string",
"pool_name": "publication_type"
},
{
"name": "publication_date",
"type": "date"
},
{
"name": "language",
"type": "string",
"pool_name": "language"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "publisher_id",
"target_entity": "publisher"
}
]
},
{
"name": "article",
"weight": 100,
"columns": [
{
"name": "article_id",
"type": "string"
},
{
"name": "publication_id",
"type": "string"
},
{
"name": "author_id",
"type": "string"
},
{
"name": "title",
"type": "string",
"pool_name": "product_names"
},
{
"name": "doi",
"type": "string"
},
{
"name": "pages",
"type": "string"
},
{
"name": "publication_date",
"type": "date"
},
{
"name": "volume",
"type": "string"
},
{
"name": "issue",
"type": "string"
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "publication_id",
"target_entity": "publication"
},
{
"source_col": "author_id",
"target_entity": "author"
}
]
},
{
"name": "review",
"weight": 200,
"columns": [
{
"name": "review_id",
"type": "string"
},
{
"name": "article_id",
"type": "string"
},
{
"name": "reviewer_id",
"type": "string"
},
{
"name": "review_date",
"type": "date"
},
{
"name": "recommendation",
"type": "string",
"pool_name": "recommendation"
},
{
"name": "comments",
"type": "string",
"pool_name": "word",
"null_rate_default": 0.4
},
{
"name": "source_system",
"type": "string"
}
],
"fk_remaps": [
{
"source_col": "article_id",
"target_entity": "article"
},
{
"source_col": "reviewer_id",
"target_entity": "author"
}
]
}
],
"hn_types": [
{
"name": "same_orcid_diff_name",
"entity_type": "author",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"orcid\"], \"attr_fields\": [\"first_name\", \"last_name\", \"affiliation\", \"email\"]}"
},
{
"name": "same_email_diff_author",
"entity_type": "author",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"email\"], \"attr_fields\": [\"first_name\", \"last_name\", \"orcid\", \"affiliation\"]}"
},
{
"name": "same_name_diff_affiliation",
"entity_type": "author",
"config_json": "{\"pattern\": \"same_name_different_everything\", \"first_name_col\": \"first_name\", \"last_name_col\": \"last_name\", \"attr_fields\": [\"affiliation\", \"orcid\", \"email\", \"h_index\"]}"
},
{
"name": "same_issn_diff_publisher",
"entity_type": "publisher",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"issn\"], \"attr_fields\": [\"publisher_name\", \"country\", \"year_founded\"]}"
},
{
"name": "same_isbn_diff_title",
"entity_type": "publication",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"isbn\"], \"attr_fields\": [\"title\", \"publication_type\", \"publication_date\", \"publisher_id\"]}"
},
{
"name": "same_title_diff_date",
"entity_type": "article",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"title\"], \"attr_fields\": [\"publication_date\", \"author_id\", \"doi\", \"pages\"]}"
},
{
"name": "same_doi_diff_publication",
"entity_type": "article",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"doi\"], \"attr_fields\": [\"publication_id\", \"title\", \"author_id\", \"pages\"]}"
},
{
"name": "same_reviewer_same_article",
"entity_type": "review",
"config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"reviewer_id\", \"article_id\"], \"attr_fields\": [\"review_date\", \"recommendation\", \"comments\"]}"
},
{
"name": "same_affiliation_diff_name",
"entity_type": "author",
"config_json": "{\"pattern\": \"mix_identifier\", \"mix_field\": \"first_name\", \"attr_fields\": [\"last_name\", \"affiliation\", \"orcid\", \"email\"]}"
}
]
}