```json { "table_name": "report_evidence", "primary_key": "jobs_id,artifact_id", "candidates": [ { "rank": 0, "partition_key": "jobs_id", "partition_key_label": "primary_identifier", "partition_key_reasoning": "jobs_id has RPUV=16.42 (116 unique values across 1,905 rows), indicating each job has ~16 artifacts on average. Top 3 coverage is 11.34%, showing reasonable distribution. Semantically, this is the primary organizing dimension — each analysis job produces a collection of evidence artifacts.", "clustering_key": null, "clustering_key_label": null, "clustering_key_reasoning": "Dimension tables typically do not require internal sorting. The artifact_id already provides uniqueness within each job, and there is no temporal or sequential ordering needed.", "world_line": null, "session_column": null, "confidence": "HIGH", "notes": "This is a dimension table storing metadata about analysis artifacts. The composite key (jobs_id, artifact_id) uniquely identifies each row. Partitioning by jobs_id groups all evidence artifacts for a given analysis job together, which is the natural access pattern. The _ingestion_timestamp, _batch_id, _source_file, and _source_system columns are all constant (100% coverage of single value), indicating this is a single-batch snapshot." } ] } ```