syscred/liar_benchmark_test.json

{
  "timestamp": "2026-02-03T11:06:54.379689",
  "dataset": "LIAR",
  "metrics": {
    "total_statements": 100,
    "successful_evaluations": 100,
    "error_count": 0,
    "error_rate": 0.0,
    "binary": {
      "accuracy": 0.57,
      "precision": 1.0,
      "recall": 0.1568627450980392,
      "f1": 0.2711864406779661,
      "confusion_matrix": [
        [
          8,
          43
        ],
        [
          0,
          49
        ]
      ]
    },
    "ternary": {
      "accuracy": 0.34,
      "macro_f1": 0.18686868686868685,
      "confusion_matrix": [
        [
          0,
          35,
          0
        ],
        [
          0,
          33,
          0
        ],
        [
          0,
          31,
          1
        ]
      ]
    },
    "classification_report": {
      "Fake": {
        "precision": 1.0,
        "recall": 0.1568627450980392,
        "f1-score": 0.2711864406779661,
        "support": 51.0
      },
      "Real": {
        "precision": 0.532608695652174,
        "recall": 1.0,
        "f1-score": 0.6950354609929078,
        "support": 49.0
      },
      "accuracy": 0.57,
      "macro avg": {
        "precision": 0.7663043478260869,
        "recall": 0.5784313725490196,
        "f1-score": 0.483110950835437,
        "support": 100.0
      },
      "weighted avg": {
        "precision": 0.7709782608695653,
        "recall": 0.57,
        "f1-score": 0.4788724606322875,
        "support": 100.0
      }
    },
    "score_distribution": {
      "mean": 0.5134000000000001,
      "min": 0.35,
      "max": 0.69,
      "median": 0.52
    },
    "per_party": {
      "republican": {
        "count": 43,
        "accuracy": 0.5581395348837209
      },
      "democrat": {
        "count": 27,
        "accuracy": 0.6296296296296297
      }
    },
    "elapsed_time": 108.7882571220398,
    "statements_per_second": 0.9192168589282478
  },
  "config": {
    "threshold": 0.5,
    "use_graphrag": true,
    "weights": {
      "source_reputation": 0.22,
      "domain_age": 0.08,
      "sentiment_neutrality": 0.13,
      "entity_presence": 0.13,
      "coherence": 0.12,
      "fact_check": 0.17,
      "graph_context": 0.15
    }
  }
}