Home / Strategies / Web Processing / FactCheckingStrategy

FactCheckingStrategy

Verifies specific claims against web page content using LLM-driven evidence extraction, confidence scoring, and multi-source cross-referencing.

Analysis Thread-Safe GPT-4 Preferred
⚙️ FactCheckingConfig.json
{
  "claims_to_verify": [
    "The project was founded in 2021",
    "It supports Python 3.11+"
  ],
  "required_sources": 3,
  "confidence_threshold": 0.85,
  "required_sources": 2
}
→
📊 Final Markdown Report
# Fact-Checking Report
Pages Analyzed: 12
## Claim: The project was founded in 2021 SUPPORTED
- [source.com/about](...)
   "Established in late 2021..."
## Claim: It supports Python 3.11+ CONTRADICTED
- [docs.io/install](...)
   "Currently only supports up to 3.10"

Configuration Parameters

Field Type Default Description
claims_to_verify * List<String> - The specific statements or facts to be verified against the crawled content.
confidence_threshold Double 0.8 Required confidence level (0.0-1.0) for a verdict to be considered definitive.
required_sources Int 3 Number of supporting sources required to reach a SUPPORTED verdict.
contradiction_threshold Int 2 Number of contradicting sources required to reach a CONTRADICTED verdict.