```json { "classifications": [ { "column": "id", "role": "identifier", "confidence": 1.0, "reason": "Primary key with unique values per row (902 distinct for 902 rows)" }, { "column": "dataset_id", "role": "identifier", "confidence": 1.0, "reason": "Foreign key referencing datasets, used for JOINs" }, { "column": "batch_id", "role": "identifier", "confidence": 1.0, "reason": "Foreign key referencing batches, used for JOINs" }, { "column": "engine", "role": "dimension", "confidence": 1.0, "reason": "Low cardinality categorical field (2 values: duckdb, whoosh) for grouping/filtering" }, { "column": "query_text", "role": "attribute", "confidence": 0.95, "reason": "High cardinality (767 distinct) free-form text content for display, not for grouping" }, { "column": "description", "role": "attribute", "confidence": 1.0, "reason": "High cardinality (824 distinct) descriptive text for display purposes" }, { "column": "model_name", "role": "dimension", "confidence": 1.0, "reason": "Low cardinality categorical field (2 values) for grouping/filtering by AI model" }, { "column": "latency_ms", "role": "measure", "confidence": 1.0, "reason": "Numeric quantity (milliseconds) to be aggregated with sum/avg/min/max" }, { "column": "created_at", "role": "dimension", "confidence": 1.0, "reason": "Timestamp for temporal grouping, filtering, and time-series analysis" } ] } ```