dupehell 0.6.0

Synthetic multi-domain dataset generator for record linkage benchmarking
Documentation
{
  "domain": "aviation",
  "entities": [
    {
      "name": "airline",
      "weight": 1,
      "columns": [
        {
          "name": "airline_id",
          "type": "string"
        },
        {
          "name": "airline_name",
          "type": "string",
          "pool_name": "company"
        },
        {
          "name": "iata_code",
          "type": "string"
        },
        {
          "name": "country_code",
          "type": "string"
        },
        {
          "name": "fleet_count",
          "type": "int"
        },
        {
          "name": "alliance",
          "type": "string",
          "pool_name": "airline_alliance"
        },
        {
          "name": "source_system",
          "type": "string"
        }
      ],
      "fk_remaps": []
    },
    {
      "name": "aircraft",
      "weight": 20,
      "columns": [
        {
          "name": "aircraft_id",
          "type": "string"
        },
        {
          "name": "airline_id",
          "type": "string"
        },
        {
          "name": "tail_number",
          "type": "string"
        },
        {
          "name": "aircraft_type",
          "type": "string",
          "pool_name": "aircraft_type"
        },
        {
          "name": "manufacturer",
          "type": "string",
          "pool_name": "company"
        },
        {
          "name": "manufacture_year",
          "type": "int"
        },
        {
          "name": "seat_capacity",
          "type": "int"
        },
        {
          "name": "source_system",
          "type": "string"
        }
      ],
      "fk_remaps": [
        {
          "source_col": "airline_id",
          "target_entity": "airline",
          "fk_weight": 1.0
        }
      ]
    },
    {
      "name": "flight",
      "weight": 300,
      "columns": [
        {
          "name": "flight_id",
          "type": "string"
        },
        {
          "name": "airline_id",
          "type": "string"
        },
        {
          "name": "aircraft_id",
          "type": "string"
        },
        {
          "name": "flight_number",
          "type": "string"
        },
        {
          "name": "origin_airport",
          "type": "string",
          "pool_name": "city"
        },
        {
          "name": "destination_airport",
          "type": "string",
          "pool_name": "city"
        },
        {
          "name": "departure_time",
          "type": "datetime"
        },
        {
          "name": "arrival_time",
          "type": "datetime"
        },
        {
          "name": "status",
          "type": "string",
          "pool_name": "status_active"
        },
        {
          "name": "source_system",
          "type": "string"
        }
      ],
      "fk_remaps": [
        {
          "source_col": "airline_id",
          "target_entity": "airline",
          "fk_weight": 1.0
        },
        {
          "source_col": "aircraft_id",
          "target_entity": "aircraft",
          "fk_weight": 0.5
        }
      ]
    },
    {
      "name": "passenger",
      "weight": 250,
      "columns": [
        {
          "name": "passenger_id",
          "type": "string"
        },
        {
          "name": "flight_id",
          "type": "string"
        },
        {
          "name": "first_name",
          "type": "string"
        },
        {
          "name": "last_name",
          "type": "string"
        },
        {
          "name": "seat_number",
          "type": "string"
        },
        {
          "name": "booking_reference",
          "type": "string"
        },
        {
          "name": "frequent_flyer",
          "type": "string",
          "null_rate_default": 0.6
        },
        {
          "name": "fare_class",
          "type": "string",
          "pool_name": "fare_class"
        },
        {
          "name": "source_system",
          "type": "string"
        }
      ],
      "fk_remaps": [
        {
          "source_col": "flight_id",
          "target_entity": "flight",
          "fk_weight": 1.0
        }
      ]
    }
  ],
  "hn_types": [
    {
      "name": "same_iata_diff_name",
      "entity_type": "airline",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"iata_code\"], \"attr_fields\": [\"airline_name\", \"country_code\", \"fleet_count\"]}"
    },
    {
      "name": "same_tail_number_diff_airline",
      "entity_type": "aircraft",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"tail_number\"], \"attr_fields\": [\"aircraft_type\", \"manufacturer\", \"manufacture_year\"]}"
    },
    {
      "name": "same_flight_number_same_route",
      "entity_type": "flight",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"flight_number\", \"origin_airport\", \"destination_airport\"], \"attr_fields\": [\"departure_time\", \"arrival_time\", \"status\"]}"
    },
    {
      "name": "same_booking_reference_diff_passenger",
      "entity_type": "passenger",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"booking_reference\"], \"attr_fields\": [\"first_name\", \"last_name\", \"seat_number\"]}"
    },
    {
      "name": "same_name_same_flight",
      "entity_type": "passenger",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"first_name\", \"last_name\", \"flight_id\"], \"attr_fields\": [\"seat_number\", \"booking_reference\", \"fare_class\"]}"
    },
    {
      "name": "same_frequent_flyer_diff_name",
      "entity_type": "passenger",
      "config_json": "{\"pattern\": \"same_field\", \"id_fields\": [\"frequent_flyer\"], \"attr_fields\": [\"first_name\", \"last_name\", \"fare_class\"]}"
    }
  ]
}