Skip to content
TruthVouch Docs

Semantic Understanding

TruthVouch uses AI to understand the semantic meaning of text, enabling intelligent matching even when questions are phrased differently. For example, “When was TruthVouch founded?” and “What’s TruthVouch’s founding year?” are recognized as the same question — enabling fast, accurate fact-checking regardless of wording.

What Semantic Understanding Powers

Hallucination Detection

TruthVouch determines the relationship between statements — whether one supports, contradicts, or is unrelated to another — using Natural Language Inference (NLI) for hallucination detection.

Fact Matching

When a new claim arrives, TruthVouch compares it against your entire knowledge base using semantic similarity, finding the most relevant truth nuggets in milliseconds. This works even when the claim uses completely different phrasing from your verified facts.

Query Recommendations

By understanding meaning, TruthVouch can suggest related queries based on what similar users have asked, helping you discover monitoring gaps.

Response Caching

Semantic understanding enables the cache to recognize when a new query is asking the same thing as a previously verified query, returning instant results.

How It Works

TruthVouch automatically processes your truth nuggets and incoming AI responses to understand their semantic meaning. When a new claim arrives, it’s compared against your knowledge base using semantic similarity. This happens in milliseconds, enabling real-time hallucination detection.

Enterprise Customization

For organizations with domain-specific terminology, TruthVouch supports custom models optimized for your industry or specialized knowledge domain. Contact your account team to configure this.

Next Steps