Knowledge & Data · Software component
Multi-Hop Retrieval Controller
Software componentKnowledge & DataKnowledge & Dataarc:MultiHopRetrievalController
A retrieval control component that answers a question requiring evidence from several documents by iterating hops: resolving each sub-question, retrieving evidence, carrying intermediate answers forward, and handing results to synthesis.
Responsibility. Sequences retrieval hops and maintains intermediate reasoning state across them.
Also known as: Multi-hop QA agent, Iterative retrieval loop
Relationships
delegates to dependency
invokes dependency
is invoked by dependency
writes dependency
- Working Memory Buffer abstract Ch3.3
is evaluated by assurance
Design guidance
- SHOULD maintain reasoning state across hops so earlier findings inform later retrieval.
- SHOULD formulate next-hop retrieval queries from intermediate results, since second-hop retrieval errors cascade.
- SHOULD detect insufficient information and decline rather than hallucinate a confident answer.
- MAY fall back to web navigation when required information is absent from retrieved documents.
Quantitative guidance
As stated by the sources; verify before use.
- Three-hop tasks show 30-50% lower success than equivalent two-hop tasks; four-hop drops further (Ch3.3).
- Reported error propagation rate 28% on minor tasks and 72% on major tasks (Ch3.3).
Classification
- Patterns
- Multi-hop retrievalBridge and comparison reasoningRecall-oriented retrieval then reasoning-based filtering
- Quality attributes
- Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Explainability (NIST AI RMF: explainable and interpretable)
- Risks mitigated
- Error propagation from early hops
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.