{
  "identity_columns": [
    "unique_id",
    "record_id",
    "source_dataset"
  ],
  "matching_columns": [
    "business_name",
    "tax_code",
    "representative_name",
    "address",
    "industry"
  ],
  "temporal_columns": [
    "record_date"
  ],
  "control_columns": [
    "data_split"
  ],
  "forbidden_columns": [
    "true_entity_id",
    "scenario_group",
    "pair_type",
    "label"
  ]
}