Explorer
/tmp/kc-v2-adversarial.json
← Zurück ↓ Download
{
  "run_id": "kc-ro-20260722T110813Z",
  "curator_version": "knowledge-curator-ro/0.2.0",
  "normalization_version": "kc-norm-v2",
  "read_only": true,
  "automatic_current_assignments": 0,
  "automatic_verified_assignments": 0,
  "adversarial_tests": {
    "total": 25,
    "passed": 25,
    "failed": 0,
    "results": [
      {
        "id": "01_no_claim",
        "text": "Nur eine Überschrift ohne Aussage.",
        "expected": {
          "candidate_type": null,
          "modality": null,
          "polarity": null,
          "review": false,
          "conflict": "none"
        },
        "actual": {
          "candidate_type": null,
          "modality": null,
          "polarity": null,
          "review": false,
          "conflict": "none"
        },
        "passed": true
      },
      {
        "id": "02_question",
        "text": "Frage: Läuft Dienst X auf Server Y?",
        "expected": {
          "candidate_type": null,
          "modality": null,
          "polarity": null,
          "review": false,
          "conflict": "none"
        },
        "actual": {
          "candidate_type": null,
          "modality": null,
          "polarity": null,
          "review": false,
          "conflict": "none"
        },
        "passed": true
      },
      {
        "id": "03_negation",
        "text": "Dienst X läuft nicht auf Server Y.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "negative",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "negative",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "04_should",
        "text": "Dienst X soll auf Server Y laufen.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "05_plan",
        "text": "Das Team plant den Providerwechsel.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "06_rejected",
        "text": "Der geplante Import wurde verworfen.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "withdrawn",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "withdrawn",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "07_rollback",
        "text": "Der Dienst wurde zurückgerollt.",
        "expected": {
          "candidate_type": "KnowledgeEvent",
          "modality": "rolled_back",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeEvent",
          "modality": "rolled_back",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "08_quote",
        "text": "Zitat: \"Dienst X läuft produktiv.\"",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "reported",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "reported",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "09_multiple",
        "text": "Dienst X läuft. Dienst Y ist aktiv.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "10_cross_chunk",
        "text": "Dienst X ist aktiv; Fortsetzung im nächsten Chunk.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "11_same_other_time",
        "text": "Dienst X war aktiv.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "12_contradiction",
        "text": "Dienst X läuft auf Server Y.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "13_alias",
        "text": "Hermes-Carlo verwendet OpenRouter.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "14_similar_subject",
        "text": "Hermes-Maria verwendet OpenRouter.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "15_version_only",
        "text": "Businessplan Version 3 wurde durch Version 4 ersetzt.",
        "expected": {
          "candidate_type": "KnowledgeEvent",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeEvent",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "16_fact_conflict",
        "text": "Businessplan Version 4 behauptet: Umsatz ist 20.000 EUR.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "17_policy",
        "text": "Policy: Ein technischer Abschluss muss einen Nachweis haben.",
        "expected": {
          "candidate_type": "KnowledgeRule",
          "modality": null,
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeRule",
          "modality": null,
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "18_capability_planned",
        "text": "Vento plant Telegram-Unterstützung.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "planned",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "19_capability_tested",
        "text": "Telegram-Unterstützung wurde getestet.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "tested",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "tested",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "20_capability_prod",
        "text": "Telegram-Unterstützung ist produktiv bestätigt.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "confirmed",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "confirmed",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "21_negative_capability",
        "text": "Vento unterstützt Telegram nicht.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "negative",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "negative",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "22_historical",
        "text": "Im Mai 2026 war Hermes auf Codex/OAuth.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "23_event",
        "text": "DeepSeek wurde durch OpenRouter ersetzt.",
        "expected": {
          "candidate_type": "KnowledgeEvent",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeEvent",
          "modality": "factual_historical",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "24_equal_sources",
        "text": "Zwei gleichwertige Quellen nennen unterschiedliche Werte.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "factual_current",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      },
      {
        "id": "25_weak_new",
        "text": "Eine Arbeitsnotiz nennt einen neuen, unbestätigten Wert.",
        "expected": {
          "candidate_type": "KnowledgeClaim",
          "modality": "reported",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "actual": {
          "candidate_type": "KnowledgeClaim",
          "modality": "reported",
          "polarity": "positive",
          "review": true,
          "conflict": "manual"
        },
        "passed": true
      }
    ]
  }
}