```json { "table_name": "deleted_datasets", "primary_key": "dataset_id", "candidates": [ { "rank": 0, "partition_key": "dataset_id", "partition_key_label": "E", "partition_key_reasoning": "Entity column identified by classifier. RPUV=1.00 (9 unique values for 9 rows), uniqueness=1.000. Each dataset_id represents a distinct dataset entity whose deletion lifecycle is being tracked. Semantic role: the primary entity whose path through creation→deletion→purge is recorded.", "clustering_key": "created_at", "clustering_key_label": "S", "clustering_key_reasoning": "Time axis identified by classifier as event_time. RPUV=1.00, uniqueness=1.000. Represents the temporal ordering of dataset lifecycle events. Using created_at as the primary sequence axis since it anchors the lifecycle timeline (created→deleted→purged).", "world_line": "dataset_id", "session_column": null, "confidence": "HIGH", "notes": "This is a path table tracking the lifecycle of deleted datasets. Each dataset_id has a unique timeline through creation, deletion, and purging. The table captures the complete audit trail of dataset removal. With only 9 rows, this is a small reference table, but the path structure is clear: partition by entity (dataset_id), order by temporal sequence (created_at). Alternative clustering by deleted_at or purged_at would also be valid depending on analytical focus, but created_at provides the canonical start of the lifecycle." } ] } ```