```json { "table_name": "classification_candidates", "table_type_tags": [ {"tag": "dimension", "confidence": 0.95} ], "primary_type": "dimension", "insufficient_evidence": false, "entity": { "columns": ["id"], "unique_per_row": true }, "time_axis": null, "reasoning": "id is unique per row (Uniqueness 1.000, RPUV 1.00). Each row is a candidate recommendation for how to partition/cluster a classified table — a catalog entry describing one storage strategy option. The table_classification_id references the parent classification (RPUV 1.15 means some classifications have multiple candidate strategies, as seen in rank 0/1/2 alternatives). The _ingestion_timestamp and _batch_id/_source_file/_source_system columns are provenance metadata, ignored per instructions. No time axis orders these recommendations — they are alternative strategies generated at classification time, not a timeline. Dimension: each candidate is unique information about one storage approach." } ```