Skip to main content

Verification API

Multi-layered content verification endpoints.

Claim Extraction​

Extract Claims from Article​

POST /api/verification/verify-claims/{contentId}

Query params:

  • force=true - Re-extract claims even if cached

Returns extracted factual claims with initial verification status.

Get Claims for Article​

GET /api/verification/claims/{contentId}

Batch Extract Recent Content​

POST /api/verification/claims/extract-recent?limit=10&minConfidence=60

Cross-Reference Verification​

Cross-Reference Single Claim​

POST /api/verification/cross-reference/claim/{claimId}?daysBack=7

Searches the article database for corroborating or contradicting sources.

Cross-Reference All Claims for Article​

POST /api/verification/cross-reference/content/{contentId}?daysBack=7

Batch Cross-Reference​

POST /api/verification/cross-reference/batch?limit=20&daysBack=7

Cross-Reference Statistics​

GET /api/verification/cross-reference/stats

Response:

{
  "success": true,
  "data": {
    "totalReferences": 1250,
    "accurateReferences": 980,
    "accuracyRate": 78,
    "topCorroboratingSources": [...],
    "groundTruthSources": ["Associated Press", "AP News", "Reuters", ...]
  }
}

Bias Detection​

Analyze Article Bias​

POST /api/verification/bias/{contentId}

Response:

{
  "success": true,
  "data": {
    "contentId": "...",
    "politicalBias": "center-left",
    "politicalConfidence": 65,
    "emotionalLevel": "low",
    "sensationalismLevel": "none",
    "indicators": {
      "loadedLanguage": [],
      "unsupportedClaims": [],
      "cherryPicking": false,
      "adHominem": false,
      "falseBalance": false,
      "omission": ["missing poll methodology"]
    },
    "overallBiasScore": 35,
    "summary": "...",
    "recommendations": [...]
  }
}

Batch Bias Analysis​

POST /api/verification/bias/batch?limit=20

Source Bias Summary​

GET /api/verification/bias/source/{sourceId}

Verification Trail​

Full Verification Trail​

GET /api/verification/trail/{contentId}

Step-by-step breakdown of confidence score calculation:

{
  "success": true,
  "data": {
    "contentId": "...",
    "contentTitle": "...",
    "finalConfidenceScore": 75,
    "confidenceLevel": "medium",
    "steps": [
      {
        "type": "source",
        "label": "Source Reliability",
        "description": "Reuters has a reliability score of 85/100...",
        "impact": "positive",
        "scoreContribution": 10
      },
      {
        "type": "ground_truth",
        "label": "Wire Service Source",
        "description": "Reuters is a wire service (ground truth source)...",
        "impact": "positive",
        "scoreContribution": 15
      }
    ],
    "summary": {
      "positiveFactors": ["High source reliability", "Wire service"],
      "negativeFactors": [],
      "recommendation": "This content appears highly reliable..."
    },
    "comparison": {
      "sourceAverage": 72,
      "domainAverage": 65,
      "percentileRank": 85
    }
  }
}

Verification Summary (Lightweight)​

GET /api/verification/summary/{contentId}

Deep Verification​

Run the full verification pipeline in one call:

POST /api/verification/deep/{contentId}

Performs:

  1. Claim extraction via LLM
  2. Cross-reference verification
  3. Bias analysis
  4. Trail generation

Response includes all results plus final recommendation.

⚠️ Note: This is expensive (multiple LLM calls). Use for high-priority content.

LLM Verification​

Verify Single Article​

POST /api/verification/llm/{contentId}

Batch LLM Verification​

POST /api/verification/llm/batch?limit=10

Heuristic Verification​

Faster, cheaper verification using pattern matching:

POST /api/verification/content/{contentId}
POST /api/verification/batch?limit=50

Statistics​

Overview​

GET /api/verification/stats/overview

Response:

{
  "success": true,
  "data": {
    "verified": 6000,
    "unverified": 30000,
    "averageConfidence": 62,
    "distribution": {
      "high": 2300,
      "medium": 3700,
      "low": 1
    }
  }
}