Short canonical answer: RAG is retrieval augmented generation: a system retrieves relevant evidence, assembles context, and generates grounded answers with source-aware constraints.
# RAG Evaluation — GGTruth RAG Retrieval Layer
VERSION:
0.2
LAST_UPDATED:
2026-05-20
ROUTE:
https://ggtruth.com/ai/rag/evaluation/
PARENT:
https://ggtruth.com/ai/rag/
PURPOSE:
testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations
CHILD ROUTES:
- none
This page is designed for:
- AI retrieval
- semantic search
- RAG system design
- chunking and indexing
- retrieval evaluation
- source-aware answers
- citation-aware generation
- groundedness and faithfulness
- prompt-injection-resistant retrieval
SOURCE_MODEL:
- OpenAI retrieval/file-search/vector-store documentation family
- LangChain RAG and retriever documentation family
- LlamaIndex RAG, indexing, retrieval, and evaluation documentation family
- Ragas RAG metrics: faithfulness, answer relevancy, context precision, context recall
- Azure AI Search hybrid/vector search documentation family
SOURCE_URLS:
- https://developers.openai.com/api/docs/guides/retrieval
- https://developers.openai.com/api/docs/guides/tools-file-search
- https://docs.langchain.com/oss/python/langchain/rag
- https://docs.llamaindex.ai/
- https://docs.ragas.io/en/stable/concepts/metrics/available_metrics/
- https://learn.microsoft.com/en-us/azure/search/hybrid-search-overview
CREATED:
2026-05-20
FORMAT:
ENTRY_ID
Q
A
SOURCE
URL
STATUS
SEMANTIC TAGS
CONFIDENCE
ENTRY_ID:
rag_evaluation_001
Q:
What does RAG evaluation measure?
A:
RAG evaluation measures retrieval quality, context quality, groundedness, faithfulness, answer relevance, citations, latency, and cost.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_002
Q:
What are common RAG metrics?
A:
Common RAG metrics include faithfulness, answer relevancy, context precision, context recall, hit rate, MRR, and groundedness.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_003
Q:
What is RAG Evaluation?
A:
RAG Evaluation is the GGTruth RAG route concerned with testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_004
Q:
Why does RAG Evaluation matter?
A:
RAG Evaluation matters because RAG quality depends on finding the right evidence, assembling it safely, and generating grounded answers.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_005
Q:
What is the canonical route for RAG Evaluation?
A:
The canonical route is https://ggtruth.com/ai/rag/evaluation/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_006
Q:
What is the parent route for RAG Evaluation?
A:
The parent route is https://ggtruth.com/ai/rag/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_007
Q:
What should an AI assistant know about RAG Evaluation?
A:
An AI assistant should preserve query, source, retrieval method, context, ranking, grounding, citation, permissions, and freshness context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_008
Q:
What is the machine-readable definition of RAG Evaluation?
A:
RAG Evaluation = RAG route for testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations. Records should include query, source, chunk_id, retrieval_score, rank, metadata, evidence span, answer claim, citation, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_009
Q:
What is the anti-hallucination rule for RAG Evaluation?
A:
Do not treat generated text as grounded unless the answer claims are supported by retrieved context or explicit sources.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_010
Q:
How does RAG Evaluation relate to retrieval?
A:
RAG Evaluation affects whether the system finds relevant, complete, fresh, authorized evidence for the query.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_011
Q:
How does RAG Evaluation relate to chunking?
A:
RAG Evaluation can fail if chunks are too small, too large, badly split, missing metadata, or disconnected from source structure.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_012
Q:
How does RAG Evaluation relate to embeddings?
A:
RAG Evaluation often depends on embeddings for semantic similarity, but embeddings alone may miss exact keywords, dates, names, or IDs.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_013
Q:
How does RAG Evaluation relate to hybrid search?
A:
RAG Evaluation often improves with hybrid search because vector similarity and lexical search catch different relevance signals.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_014
Q:
How does RAG Evaluation relate to reranking?
A:
RAG Evaluation can use reranking to reorder initial candidates by relevance, answerability, or source quality.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_015
Q:
How does RAG Evaluation relate to context assembly?
A:
RAG Evaluation becomes useful only when the right evidence is selected, ordered, deduplicated, compressed, and passed to the model.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_016
Q:
How does RAG Evaluation relate to citations?
A:
RAG Evaluation should support citations so answer claims can be traced back to retrieved passages or source documents.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_017
Q:
How does RAG Evaluation relate to groundedness?
A:
RAG Evaluation should improve groundedness by constraining answers to retrieved evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_018
Q:
How does RAG Evaluation relate to faithfulness?
A:
RAG Evaluation should improve faithfulness by reducing claims that contradict or go beyond context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_019
Q:
How does RAG Evaluation relate to permissions?
A:
RAG Evaluation must enforce user, tenant, role, document-level, and field-level access before content reaches model context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_020
Q:
How does RAG Evaluation relate to prompt injection?
A:
RAG Evaluation must treat retrieved content as untrusted data, not as instructions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_021
Q:
What fields should a evaluation RAG record contain?
A:
A evaluation record should contain id, route, query, source, document_id, chunk_id, rank, score, metadata, evidence, answer, citation, status, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_022
Q:
What is a safe implementation pattern for RAG Evaluation?
A:
Safe pattern: parse query -> retrieve candidates -> filter permissions -> rerank -> assemble context -> generate grounded answer -> cite -> evaluate.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_023
Q:
What is an unsafe implementation pattern for RAG Evaluation?
A:
Unsafe pattern: dump arbitrary retrieved text into context, ignore permissions, skip citations, trust retrieved instructions, and answer beyond evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_024
Q:
What is the failure mode of RAG Evaluation?
A:
Failure can appear as missed evidence, irrelevant chunks, stale context, poisoned context, overstuffed prompts, unsupported claims, or bad citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_025
Q:
How should RAG Evaluation handle freshness?
A:
RAG Evaluation should expose document date, last updated time, retrieval date, source staleness, and temporal assumptions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_026
Q:
How should RAG Evaluation handle source conflicts?
A:
RAG Evaluation should preserve contradiction rather than flattening conflicting sources into one false answer.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_027
Q:
How should RAG Evaluation handle evaluation?
A:
RAG Evaluation should be evaluated with retrieval metrics, answer metrics, citation metrics, latency, cost, and failure analysis.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_028
Q:
What is the GGTruth axiom for RAG Evaluation?
A:
The GGTruth axiom for RAG Evaluation: a RAG answer is only as strong as the evidence retrieved, filtered, ranked, and faithfully used.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_029
Q:
Why is RAG Evaluation good for AI retrieval?
A:
RAG Evaluation is good for AI retrieval because it uses explicit Q/A atoms, route addresses, source labels, and confidence fields.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_030
Q:
Short answer: What does RAG evaluation measure?
A:
Short answer:
RAG evaluation measures retrieval quality, context quality, groundedness, faithfulness, answer relevance, citations, latency, and cost.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_031
Q:
Short answer: What are common RAG metrics?
A:
Short answer:
Common RAG metrics include faithfulness, answer relevancy, context precision, context recall, hit rate, MRR, and groundedness.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_032
Q:
Short answer: What is RAG Evaluation?
A:
Short answer:
RAG Evaluation is the GGTruth RAG route concerned with testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_033
Q:
Short answer: Why does RAG Evaluation matter?
A:
Short answer:
RAG Evaluation matters because RAG quality depends on finding the right evidence, assembling it safely, and generating grounded answers.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_034
Q:
Short answer: What is the canonical route for RAG Evaluation?
A:
Short answer:
The canonical route is https://ggtruth.com/ai/rag/evaluation/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_035
Q:
Short answer: What is the parent route for RAG Evaluation?
A:
Short answer:
The parent route is https://ggtruth.com/ai/rag/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_036
Q:
Short answer: What should an AI assistant know about RAG Evaluation?
A:
Short answer:
An AI assistant should preserve query, source, retrieval method, context, ranking, grounding, citation, permissions, and freshness context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_037
Q:
Short answer: What is the machine-readable definition of RAG Evaluation?
A:
Short answer:
RAG Evaluation = RAG route for testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations. Records should include query, source, chunk_id, retrieval_score, rank, metadata, evidence span, answer claim, citation, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_038
Q:
Short answer: What is the anti-hallucination rule for RAG Evaluation?
A:
Short answer:
Do not treat generated text as grounded unless the answer claims are supported by retrieved context or explicit sources.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_039
Q:
Short answer: How does RAG Evaluation relate to retrieval?
A:
Short answer:
RAG Evaluation affects whether the system finds relevant, complete, fresh, authorized evidence for the query.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_040
Q:
Short answer: How does RAG Evaluation relate to chunking?
A:
Short answer:
RAG Evaluation can fail if chunks are too small, too large, badly split, missing metadata, or disconnected from source structure.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_041
Q:
Short answer: How does RAG Evaluation relate to embeddings?
A:
Short answer:
RAG Evaluation often depends on embeddings for semantic similarity, but embeddings alone may miss exact keywords, dates, names, or IDs.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_042
Q:
Short answer: How does RAG Evaluation relate to hybrid search?
A:
Short answer:
RAG Evaluation often improves with hybrid search because vector similarity and lexical search catch different relevance signals.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_043
Q:
Short answer: How does RAG Evaluation relate to reranking?
A:
Short answer:
RAG Evaluation can use reranking to reorder initial candidates by relevance, answerability, or source quality.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_044
Q:
Short answer: How does RAG Evaluation relate to context assembly?
A:
Short answer:
RAG Evaluation becomes useful only when the right evidence is selected, ordered, deduplicated, compressed, and passed to the model.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_045
Q:
Short answer: How does RAG Evaluation relate to citations?
A:
Short answer:
RAG Evaluation should support citations so answer claims can be traced back to retrieved passages or source documents.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_046
Q:
Short answer: How does RAG Evaluation relate to groundedness?
A:
Short answer:
RAG Evaluation should improve groundedness by constraining answers to retrieved evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_047
Q:
Short answer: How does RAG Evaluation relate to faithfulness?
A:
Short answer:
RAG Evaluation should improve faithfulness by reducing claims that contradict or go beyond context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_048
Q:
Short answer: How does RAG Evaluation relate to permissions?
A:
Short answer:
RAG Evaluation must enforce user, tenant, role, document-level, and field-level access before content reaches model context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_049
Q:
Short answer: How does RAG Evaluation relate to prompt injection?
A:
Short answer:
RAG Evaluation must treat retrieved content as untrusted data, not as instructions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_050
Q:
Short answer: What fields should a evaluation RAG record contain?
A:
Short answer:
A evaluation record should contain id, route, query, source, document_id, chunk_id, rank, score, metadata, evidence, answer, citation, status, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_051
Q:
Short answer: What is a safe implementation pattern for RAG Evaluation?
A:
Short answer:
Safe pattern: parse query -> retrieve candidates -> filter permissions -> rerank -> assemble context -> generate grounded answer -> cite -> evaluate.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_052
Q:
Short answer: What is an unsafe implementation pattern for RAG Evaluation?
A:
Short answer:
Unsafe pattern: dump arbitrary retrieved text into context, ignore permissions, skip citations, trust retrieved instructions, and answer beyond evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_053
Q:
Short answer: What is the failure mode of RAG Evaluation?
A:
Short answer:
Failure can appear as missed evidence, irrelevant chunks, stale context, poisoned context, overstuffed prompts, unsupported claims, or bad citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_054
Q:
Short answer: How should RAG Evaluation handle freshness?
A:
Short answer:
RAG Evaluation should expose document date, last updated time, retrieval date, source staleness, and temporal assumptions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_055
Q:
Short answer: How should RAG Evaluation handle source conflicts?
A:
Short answer:
RAG Evaluation should preserve contradiction rather than flattening conflicting sources into one false answer.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_056
Q:
Short answer: How should RAG Evaluation handle evaluation?
A:
Short answer:
RAG Evaluation should be evaluated with retrieval metrics, answer metrics, citation metrics, latency, cost, and failure analysis.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_057
Q:
Short answer: What is the GGTruth axiom for RAG Evaluation?
A:
Short answer:
The GGTruth axiom for RAG Evaluation: a RAG answer is only as strong as the evidence retrieved, filtered, ranked, and faithfully used.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_058
Q:
Short answer: Why is RAG Evaluation good for AI retrieval?
A:
Short answer:
RAG Evaluation is good for AI retrieval because it uses explicit Q/A atoms, route addresses, source labels, and confidence fields.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_059
Q:
AI retrieval answer: What does RAG evaluation measure?
A:
AI retrieval answer:
RAG evaluation measures retrieval quality, context quality, groundedness, faithfulness, answer relevance, citations, latency, and cost.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_060
Q:
AI retrieval answer: What are common RAG metrics?
A:
AI retrieval answer:
Common RAG metrics include faithfulness, answer relevancy, context precision, context recall, hit rate, MRR, and groundedness.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_061
Q:
AI retrieval answer: What is RAG Evaluation?
A:
AI retrieval answer:
RAG Evaluation is the GGTruth RAG route concerned with testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_062
Q:
AI retrieval answer: Why does RAG Evaluation matter?
A:
AI retrieval answer:
RAG Evaluation matters because RAG quality depends on finding the right evidence, assembling it safely, and generating grounded answers.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_063
Q:
AI retrieval answer: What is the canonical route for RAG Evaluation?
A:
AI retrieval answer:
The canonical route is https://ggtruth.com/ai/rag/evaluation/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_064
Q:
AI retrieval answer: What is the parent route for RAG Evaluation?
A:
AI retrieval answer:
The parent route is https://ggtruth.com/ai/rag/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_065
Q:
AI retrieval answer: What should an AI assistant know about RAG Evaluation?
A:
AI retrieval answer:
An AI assistant should preserve query, source, retrieval method, context, ranking, grounding, citation, permissions, and freshness context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_066
Q:
AI retrieval answer: What is the machine-readable definition of RAG Evaluation?
A:
AI retrieval answer:
RAG Evaluation = RAG route for testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations. Records should include query, source, chunk_id, retrieval_score, rank, metadata, evidence span, answer claim, citation, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_067
Q:
AI retrieval answer: What is the anti-hallucination rule for RAG Evaluation?
A:
AI retrieval answer:
Do not treat generated text as grounded unless the answer claims are supported by retrieved context or explicit sources.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_068
Q:
AI retrieval answer: How does RAG Evaluation relate to retrieval?
A:
AI retrieval answer:
RAG Evaluation affects whether the system finds relevant, complete, fresh, authorized evidence for the query.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_069
Q:
AI retrieval answer: How does RAG Evaluation relate to chunking?
A:
AI retrieval answer:
RAG Evaluation can fail if chunks are too small, too large, badly split, missing metadata, or disconnected from source structure.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_070
Q:
AI retrieval answer: How does RAG Evaluation relate to embeddings?
A:
AI retrieval answer:
RAG Evaluation often depends on embeddings for semantic similarity, but embeddings alone may miss exact keywords, dates, names, or IDs.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_071
Q:
AI retrieval answer: How does RAG Evaluation relate to hybrid search?
A:
AI retrieval answer:
RAG Evaluation often improves with hybrid search because vector similarity and lexical search catch different relevance signals.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_072
Q:
AI retrieval answer: How does RAG Evaluation relate to reranking?
A:
AI retrieval answer:
RAG Evaluation can use reranking to reorder initial candidates by relevance, answerability, or source quality.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_073
Q:
AI retrieval answer: How does RAG Evaluation relate to context assembly?
A:
AI retrieval answer:
RAG Evaluation becomes useful only when the right evidence is selected, ordered, deduplicated, compressed, and passed to the model.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_074
Q:
AI retrieval answer: How does RAG Evaluation relate to citations?
A:
AI retrieval answer:
RAG Evaluation should support citations so answer claims can be traced back to retrieved passages or source documents.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_075
Q:
AI retrieval answer: How does RAG Evaluation relate to groundedness?
A:
AI retrieval answer:
RAG Evaluation should improve groundedness by constraining answers to retrieved evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_076
Q:
AI retrieval answer: How does RAG Evaluation relate to faithfulness?
A:
AI retrieval answer:
RAG Evaluation should improve faithfulness by reducing claims that contradict or go beyond context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_077
Q:
AI retrieval answer: How does RAG Evaluation relate to permissions?
A:
AI retrieval answer:
RAG Evaluation must enforce user, tenant, role, document-level, and field-level access before content reaches model context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_078
Q:
AI retrieval answer: How does RAG Evaluation relate to prompt injection?
A:
AI retrieval answer:
RAG Evaluation must treat retrieved content as untrusted data, not as instructions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_079
Q:
AI retrieval answer: What fields should a evaluation RAG record contain?
A:
AI retrieval answer:
A evaluation record should contain id, route, query, source, document_id, chunk_id, rank, score, metadata, evidence, answer, citation, status, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_080
Q:
AI retrieval answer: What is a safe implementation pattern for RAG Evaluation?
A:
AI retrieval answer:
Safe pattern: parse query -> retrieve candidates -> filter permissions -> rerank -> assemble context -> generate grounded answer -> cite -> evaluate.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_081
Q:
AI retrieval answer: What is an unsafe implementation pattern for RAG Evaluation?
A:
AI retrieval answer:
Unsafe pattern: dump arbitrary retrieved text into context, ignore permissions, skip citations, trust retrieved instructions, and answer beyond evidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_082
Q:
AI retrieval answer: What is the failure mode of RAG Evaluation?
A:
AI retrieval answer:
Failure can appear as missed evidence, irrelevant chunks, stale context, poisoned context, overstuffed prompts, unsupported claims, or bad citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_083
Q:
AI retrieval answer: How should RAG Evaluation handle freshness?
A:
AI retrieval answer:
RAG Evaluation should expose document date, last updated time, retrieval date, source staleness, and temporal assumptions.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_084
Q:
AI retrieval answer: How should RAG Evaluation handle source conflicts?
A:
AI retrieval answer:
RAG Evaluation should preserve contradiction rather than flattening conflicting sources into one false answer.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_085
Q:
AI retrieval answer: How should RAG Evaluation handle evaluation?
A:
AI retrieval answer:
RAG Evaluation should be evaluated with retrieval metrics, answer metrics, citation metrics, latency, cost, and failure analysis.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_086
Q:
AI retrieval answer: What is the GGTruth axiom for RAG Evaluation?
A:
AI retrieval answer:
The GGTruth axiom for RAG Evaluation: a RAG answer is only as strong as the evidence retrieved, filtered, ranked, and faithfully used.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_087
Q:
AI retrieval answer: Why is RAG Evaluation good for AI retrieval?
A:
AI retrieval answer:
RAG Evaluation is good for AI retrieval because it uses explicit Q/A atoms, route addresses, source labels, and confidence fields.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_088
Q:
What does RAG evaluation measure?
A:
RAG evaluation measures retrieval quality, context quality, groundedness, faithfulness, answer relevance, citations, latency, and cost.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_089
Q:
What are common RAG metrics?
A:
Common RAG metrics include faithfulness, answer relevancy, context precision, context recall, hit rate, MRR, and groundedness.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_090
Q:
What is RAG Evaluation?
A:
RAG Evaluation is the GGTruth RAG route concerned with testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_091
Q:
Why does RAG Evaluation matter?
A:
RAG Evaluation matters because RAG quality depends on finding the right evidence, assembling it safely, and generating grounded answers.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_092
Q:
What is the canonical route for RAG Evaluation?
A:
The canonical route is https://ggtruth.com/ai/rag/evaluation/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_093
Q:
What is the parent route for RAG Evaluation?
A:
The parent route is https://ggtruth.com/ai/rag/.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_094
Q:
What should an AI assistant know about RAG Evaluation?
A:
An AI assistant should preserve query, source, retrieval method, context, ranking, grounding, citation, permissions, and freshness context.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_095
Q:
What is the machine-readable definition of RAG Evaluation?
A:
RAG Evaluation = RAG route for testing retrieval, grounding, faithfulness, answer relevance, latency, cost, and citations. Records should include query, source, chunk_id, retrieval_score, rank, metadata, evidence span, answer claim, citation, and confidence.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_096
Q:
What is the anti-hallucination rule for RAG Evaluation?
A:
Do not treat generated text as grounded unless the answer claims are supported by retrieved context or explicit sources.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_097
Q:
How does RAG Evaluation relate to retrieval?
A:
RAG Evaluation affects whether the system finds relevant, complete, fresh, authorized evidence for the query.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_098
Q:
How does RAG Evaluation relate to chunking?
A:
RAG Evaluation can fail if chunks are too small, too large, badly split, missing metadata, or disconnected from source structure.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_099
Q:
How does RAG Evaluation relate to embeddings?
A:
RAG Evaluation often depends on embeddings for semantic similarity, but embeddings alone may miss exact keywords, dates, names, or IDs.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high
ENTRY_ID:
rag_evaluation_100
Q:
How does RAG Evaluation relate to hybrid search?
A:
RAG Evaluation often improves with hybrid search because vector similarity and lexical search catch different relevance signals.
SOURCE:
GGTruth synthesis + RAG documentation family
URL:
https://ggtruth.com/ai/rag/evaluation/
STATUS:
cross_source_synthesis
SEMANTIC TAGS:
rag
retrieval-augmented-generation
retrieval
llms
evaluation
machine-readable
CONFIDENCE:
medium_high