dataset.json

{
  "artifacts": {
    "preferred_format": "parquet"
  },
  "dataset_id": "conceptnet",
  "standardization": {
    "canonical_column_map": {
      "end": "object",
      "relation": "predicate",
      "start": "subject"
    },
    "split_aliases": {
      "dev": "validation",
      "val": "validation",
      "valid": "validation"
    }
  }
}