```json { "classifications": [ {"column": "id", "role": "identifier", "confidence": 1.0, "reason": "Primary key with unique BIGINT values per row"}, {"column": "dataset_id", "role": "identifier", "confidence": 1.0, "reason": "Foreign key referencing dataset, used for joins"}, {"column": "batch_id", "role": "identifier", "confidence": 1.0, "reason": "Foreign key referencing batch, used for joins"}, {"column": "table_name", "role": "attribute", "confidence": 0.95, "reason": "Descriptive name for display, very high cardinality (18203 distinct)"}, {"column": "iceberg_namespace", "role": "dimension", "confidence": 1.0, "reason": "Low cardinality categorical field (2 values: api, gold) for grouping"}, {"column": "iceberg_snapshot_id", "role": "identifier", "confidence": 0.9, "reason": "Snapshot identifier for versioning, used for joins"}, {"column": "committed_at", "role": "dimension", "confidence": 1.0, "reason": "Timestamp for filtering and time-based grouping"}, {"column": "source_format", "role": "dimension", "confidence": 1.0, "reason": "Low cardinality categorical (3 values: csv, parquet, table) for grouping"}, {"column": "source_path", "role": "attribute", "confidence": 1.0, "reason": "File path for display, not for grouping or aggregation"}, {"column": "row_count", "role": "measure", "confidence": 1.0, "reason": "Numeric quantity to aggregate (sum/avg)"}, {"column": "byte_count", "role": "measure", "confidence": 1.0, "reason": "Numeric quantity to aggregate (sum/avg)"}, {"column": "column_count", "role": "measure", "confidence": 1.0, "reason": "Numeric quantity to aggregate (sum/avg)"}, {"column": "null_count_total", "role": "measure", "confidence": 1.0, "reason": "Numeric quantity to aggregate (sum/avg)"}, {"column": "schema_json", "role": "attribute", "confidence": 1.0, "reason": "Long JSON text for display, not for grouping or aggregation"}, {"column": "query", "role": "attribute", "confidence": 1.0, "reason": "Long SQL text for display, not for grouping or aggregation"}, {"column": "description", "role": "attribute", "confidence": 1.0, "reason": "Descriptive text for display, not for grouping or aggregation"}, {"column": "ingest_started_at", "role": "dimension", "confidence": 1.0, "reason": "Timestamp for filtering and time-based grouping"}, {"column": "ingest_elapsed_ms", "role": "measure", "confidence": 1.0, "reason": "Numeric duration to aggregate (sum/avg)"} ] } ```