GOURMET v31.4: Measurement Integrity Audit Report

Generated: 2026-06-11T05:34:56.216599+00:00 Audit time: 2.3s

1. Database Integrity

Detailed Table Counts

causal_signals_v23.db:

• alerts: 14 rows
• sqlite_sequence: 1 rows

gematria_kg.db:

• gematria_search_fts: 402 rows
• gematria_entities: 402 rows
• entity_embeddings: 402 rows
• analysis_metrics: 4 rows
• sqlite_sequence: 3 rows
• gematria_search_fts_data: 2 rows
• overnight_research_logs: 1 rows
• gematria_search_fts_config: 1 rows
• gematria_search_fts_idx: 0 rows
• gematria_search_fts_docsize: 0 rows
• gematria_relationships: 0 rows

gnn_link_prediction.db:

• gnn_predictions: 100 rows
• gnn_training_history: 20 rows
• sqlite_sequence: 2 rows

gourmet.db:

• location_cache: 412 rows
• domain_observations: 150 rows
• resonance_events: 107 rows
• domain_observations_live: 91 rows
• causal_signals: 69 rows
• temporal_heatmaps: 56 rows
• temporal_anomalies: 40 rows
• signal_accuracy_history: 28 rows
• sqlite_sequence: 15 rows
• causal_edges: 8 rows
• domains: 5 rows
• temporal_patterns: 5 rows
• political_indicators: 5 rows
• entity_mapping_registry: 3 rows
• domain_expansion_analysis: 2 rows
• signal_generation_log: 1 rows
• indicator_catalog_log: 1 rows
• indicator_measurement_registry: 1 rows
• pathway_mapping_log: 1 rows
• bridge_activity_logs: 1 rows
• alert_routing: 0 rows
• schema_migrations: 0 rows
• causal_graph_snapshots: 0 rows
• multi_domain_events: 0 rows
• ensemble_causal_features: 0 rows
• financial_rates: 0 rows
• social_monitor: 0 rows
• political_sources_registry: 0 rows
• political_indicators_catalog: 0 rows
• political_sources_metadata: 0 rows
• signal_translations: 0 rows

gourmet_knowledge_graph.db:

• kg_edges: 14504 rows
• kg_nodes: 10467 rows
• sqlite_sequence: 2 rows
• graph_snapshots: 1 rows
• entity_resolutions: 0 rows

2. Metric Saturation Audit

Known Saturation Issue (v30 Discovery)

σ=0.0000 walk-forward stability metric was saturated. With 42 windows and 8-14 day activation zones, at least 2 windows are active on ~100% of every day. The metric couldn’t detect instability even if it existed.

Corrected value: σ=0.1184 across 9 out-of-sample folds (using non-saturated CRITICAL-day rate metric).

Current Metric Status

Embedding Database:

• Total embeddings: 402
• Duplicate node_ids: 0 ✅ Clean
• NULL embeddings: 0 ✅ Clean
• Model distribution: []

Gematria Value Distribution (sample of 200 entities):

• Mean: 46.3, Std: 30.1
• Range: 7 - 169
• Unique values: 88

Knowledge Graph Quality:

• Isolated entities (no edges): 15 ✅
• Entities without domain: 9287 ⚠️

Digital Root Distribution:

• Unique values: 9
• Distribution: {‘1’: 21, ‘2’: 28, ‘3’: 20, ‘4’: 20, ‘5’: 27, ‘6’: 20, ‘7’: 22, ‘8’: 19, ‘9’: 23}

3. Edge-of-Window Overprediction

Status: Documented but not yet corrected

Plateau core is well-calibrated; decay shoulders (window edges) are slightly overconfident.

Recommendation: Apply decay-shoulder correction factor in v31.2 temporal embedding pipeline.

4. Findings Summary

Clean

• causal_signals_v23.db: integrity OK, 12KB
• gematria_kg.db: integrity OK, 932KB
• gnn_link_prediction.db: integrity OK, 52KB
• gourmet.db: integrity OK, 696KB
• gourmet_knowledge_graph.db: integrity OK, 6068KB
• Embeddings: no duplicates, no NULLs
• KG: only 15 isolated entities

Issues Found

• domain_observations.db: EMPTY
• gematria.db: EMPTY

5. Recommendations

1. Fix edge-of-window overprediction in v31.2 temporal embedding pipeline
2. Document measurement methodology for each primary metric
3. Add saturation detection to daily pipeline — flag any metric returning identical values for 3+ consecutive days

v31.4 Stream D — Measurement Integrity Audit. 2026-06-11T05:34:56.216678+00:00