Knowledge & Data · Software component

Supporting Fact Extractor

Software componentKnowledge & DataKnowledge & Dataarc:SupportingFactExtractor

A component that isolates the specific sentences within retrieved passages that support an answer, rather than passing whole passages to reasoning.

Responsibility. Identifies sentence-level supporting facts in retrieved passages.

Also known as: Fact-level extraction, Supporting fact identification

Direct neighbourhood (hover for relationship types)

Relationships

is invoked by dependency

Multi-Hop Retrieval Controller Ch3.3

sends data to dynamic

Multi-Hop Answer Synthesizer Ch3.3

Design guidance

SHOULD be measured with fact-level (sentence-level) metrics, not passage-level proxies.

Quantitative guidance

As stated by the sources; verify before use.

Fact-level F1 typically runs 10-20 percentage points below passage-level F1 on identical questions (Ch3.3).

Classification

Patterns: Sentence-level evidence selection
Quality attributes: Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Explainability (NIST AI RMF: explainable and interpretable)
Risks mitigated: Noise in long passages diluting attentionCorrect answers without valid supporting evidence

Sources

Ch3.3: T. Nguyen, "Web Navigation and Interaction Benchmarks," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.3. ISBN: 9798244538229.