```json { "table_name": "report_insights", "table_type_tags": [ {"tag": "event", "confidence": 0.75}, {"tag": "dimension", "confidence": 0.30} ], "primary_type": "event", "insufficient_evidence": false, "entity": { "columns": ["jobs_id"], "unique_per_row": false }, "time_axis": { "columns": ["index"], "kind": "sequence" }, "reasoning": "The entity jobs_id repeats (RPUV 11.91, Uniqueness 0.084) along the sequence axis index (Monotonicity 0.917, values 0–67). Each row records an insight message generated during a job run — table_count, column_range, or other findings — ordered by index within the job. The rows are occurrences (insights emitted) rather than state transitions: replaying them does not reconstruct a meaningful job state, just accumulates findings. Event is primary. The weak dimension signal comes from the fact that each (jobs_id, index) combination appears unique, but the sequence ordering and occurrence nature dominate." } ```