Cognition · Software component

Semantic Entropy Detector

Software componentCognitionCognition & Memoryarc:SemanticEntropyDetector

A hallucination detector that samples several responses to the same prompt and quantifies their semantic divergence, treating high variability as evidence the model is guessing.

Responsibility. Detects knowledge gaps via semantic inconsistency across sampled responses.

Also known as: Semantic consistency checking, Self-consistency sampling detector

Direct neighbourhood (hover for relationship types)

Relationships

invokes dependency

LLM Inference Service Ch3.10

evaluates assurance

Answer Synthesizer abstract Ch3.10

Design guidance

MAY sample ~10 generations per query and flag divergent factual content as likely hallucination.

Quantitative guidance

As stated by the sources; verify before use.

Semantic-entropy checking achieves ~79% hallucination detection accuracy in controlled experiments (Ch3.10).

Classification

Patterns: Semantic entropy
Quality attributes: Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)
Risks mitigated: Undetected knowledge-gap hallucinations

Sources

Ch3.10: T. Nguyen, "Efficiency Metrics," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.10. ISBN: 9798244538229.