---
title: "RAG patterns (KO)"
source: "https://systems-analysis.info/int/RAG_patterns_(KO)"
wiki: "systems-analysis.info/int"
article: "RAG_patterns_(KO)"
language: "ko"
categories:
  - "Category:Korean"
  - "Category:Large language models"
  - "Category:Prompt engineering"
revision_id: 6111
wiki_created_at: 2026-09-06T23:58:29Z
wiki_modified_at: 2026-09-06T23:58:29Z
downloaded_at: 2026-09-07T23:12:05Z
---

# RAG patterns (KO)

**RAG 패턴** (영어: *RAG Patterns*) — **Retrieval-Augmented Generation** (RAG) 시스템을 구축하기 위한 아키텍처적·방법론적 접근법의 집합이다. 이 패턴들은 환각(hallucination), 지식 노후화, 도메인 특수성 부족 등 대형 언어 모델(LLM)의 근본적인 문제를 해결하기 위해 LLM을 외부의 동적으로 접근 가능한 데이터 소스와 통합하는 방식으로 설계되었다<sup>[\[1\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-lewis2020-1)</sup>. RAG의 발전은 단순한 선형 파이프라인에서 복잡한 모듈형·에이전트형 시스템으로 이어졌다<sup>[\[2\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-survey2024-2)</sup>.

## RAG의 주요 패턴

기술이 발전함에 따라 다양한 RAG 패턴이 등장하였으며, 각 패턴은 특정 과제를 해결하고 품질·속도·비용 사이에서 고유한 절충점을 가진다.

- **Classic RAG (클래식 RAG)** — 기본 접근법으로, 사용자 쿼리를 벡터화하여 벡터 DB에서 관련 청크(chunk)를 검색하고, 찾은 청크를 질문과 함께 LLM에 입력하여 답변을 생성한다<sup>[\[1\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-lewis2020-1)</sup>.

<!-- -->

- **Multi‑Query RAG (멀티 쿼리)** — LLM이 원본 쿼리의 여러 변형·구체화된 버전을 생성하고, 모든 버전에 대해 검색을 수행한 뒤 결과를 통합함으로써 재현율(*recall*)을 높인다<sup>[\[3\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-langchain-multiquery-3)</sup>.

<!-- -->

- **HyDE (Hypothetical Document Expansion)** — 짧은 쿼리와 긴 문서 사이의 "의미론적 간극"을 극복하기 위해 사용된다. LLM이 먼저 "가상의" 답변 문서를 생성하고, 그 embedding을 검색에 활용하여 검색 품질을 향상시킨다<sup>[\[4\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-hyde-4)</sup>.

<!-- -->

- **Hybrid Retrieval (하이브리드 검색)** — 의미론적(벡터) 검색과 어휘적(BM25) 검색의 조합이다. 하이브리드 방식은 프로덕션 시스템의 표준이 되었다. 벡터 검색은 의미적 유사성을, BM25는 정확한 용어·ID·약어를 찾아내며, 결과는 퓨전(fusion)으로 통합된다<sup>[\[5\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-weaviate-hybrid-5)[\[6\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-qdrant-hybrid-6)[\[7\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-milvus-fulltext-7)</sup>.

<!-- -->

- **Re‑ranking (재순위화)** — 2단계 프로세스로, 빠른 리트리버가 후보 집합(예: 상위 100개)을 반환한 후, 크로스 인코더(또는 다른 재순위화기)가 관련성을 재계산하여 LLM에 전달할 최적의 결과(예: 상위 5개)를 선별한다<sup>[\[8\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-nogueira2019-8)[\[9\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-cohere-rerank-9)</sup>.

<!-- -->

- **Query Routing (쿼리 라우팅)** — 여러 이질적인 데이터 소스(서로 다른 인덱스·DB·API)가 있는 시스템에서, 라우터(LLM 선택기 또는 분류기)를 이용해 쿼리를 최적의 소스로 전달하며, fallback 전략도 포함한다<sup>[\[10\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-llama-router-10)</sup>.

<!-- -->

- **Agentic/Web RAG (에이전트 RAG)** — LLM이 에이전트로서 복잡한 질문을 분해하고, 반복을 계획하며, 피드백을 통해 도구(벡터 검색, 웹 검색)를 활용한다. 대표적인 구현 방식은 ReAct 패러다임<sup>[\[11\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-react-11)</sup>이며, 웹 중심의 수집과 의무적 인용에 대해서는 WebGPT<sup>[\[12\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-webgpt-12)</sup>를 참조한다.

### 인접 및 신흥 패러다임

- **GraphRAG (그래프 RAG)** — 지식 그래프를 컨텍스트 소스 및 선택 메커니즘으로 활용한다. 개체 간 연결 구조와 텍스트를 함께 검색하여 해석 가능성을 높이고 멀티홉(multi‑hop) 질문에서의 품질을 향상시킨다<sup>[\[13\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-graphrag-13)[\[14\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-graphrag-project-14)</sup>.
- **MM‑RAG (멀티모달 RAG)** — 텍스트와 시각적 소스(스캔·다이어그램·표)를 함께 처리한다. 예시로 VisRAG는 멀티모달 문서에 대한 VLM 기반 검색 및 생성을 시연한다<sup>[\[15\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-visrag-15)</sup>.
- **Packaging & Context Handling (컨텍스트 패키징)** — 검색된 청크를 prompt에 통합하는 방법: *Stuff*, *Map‑Reduce*, *Refine*, *Tree‑of‑Chunks (RAPTOR)*<sup>[\[16\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-raptor-16)</sup>.

## 패턴 비교표

| 패턴                 | 적용 시점                                            | 품질에 미치는 영향                                                                                                                                                                                                                                                                                             | 비용 / 지연 | 위험 및 한계                               |
|----------------------|------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------|--------------------------------------------|
| **Classic RAG**      | 동질적인 데이터베이스에 대한 PoC 및 간단한 Q&A       | 기본 수준; embedding 품질에 크게 의존<sup>[\[1\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-lewis2020-1)</sup>                                                                                                                                                                             | 낮음        | 표현 방식에 민감; 비관련 컨텍스트 위험     |
| **Hybrid Retrieval** | 대부분의 프로덕션 시나리오; 코드·약어·ID가 많은 경우 | 재현율 향상; 정확한 용어 커버<sup>[\[5\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-weaviate-hybrid-5)[\[6\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-qdrant-hybrid-6)[\[7\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-milvus-fulltext-7)</sup> | 낮음/중간   | 퓨전 가중치 조정 필요; 두 개의 인덱스 유지 |
| **Re‑ranking**       | 높은 정밀도가 중요한 경우                            | 상위 k에서 precision 상당히 향상<sup>[\[8\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-nogueira2019-8)[\[9\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-cohere-rerank-9)</sup>                                                                                         | 중간/높음   | 추가 지연/비용 발생                        |
| **Multi‑Query**      | 짧거나 다면적인 쿼리                                 | recall 향상<sup>[\[3\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-langchain-multiquery-3)</sup>                                                                                                                                                                                            | 중간        | 중복되거나 노이즈가 많은 변형 생성 가능    |
| **HyDE**             | "의미론적 간극"이 큰 짧거나 모호한 쿼리              | *zero‑shot* 검색 품질 향상<sup>[\[4\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-hyde-4)</sup>                                                                                                                                                                                             | 중간        | "가상의" 텍스트 품질에 의존                |
| **Query Routing**    | 여러 소스(문서 DB, SQL, API, 웹)                     | 올바른 소스 선택으로 관련성 향상<sup>[\[10\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-llama-router-10)</sup>                                                                                                                                                                             | 중간        | 라우팅 오류 = 검색 실패                    |
| **Agentic/Web RAG**  | 복잡하고 탐색적인 다단계 쿼리                        | 선형 파이프라인 범위를 넘어선 과제 해결<sup>[\[11\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-react-11)[\[12\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-webgpt-12)</sup>                                                                                            | 높음        | 복잡성, 루프 위험; 가드레일 필요           |

주요 RAG 패턴 비교

## 실제 구현 및 아키텍처

### 도입 단계

1.  **Proof of Concept (PoC):** 제한적이지만 대표성 있는 데이터셋으로 **Classic RAG**를 시작하여 embedding 품질과 기본 검색을 검증한다<sup>[\[1\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-lewis2020-1)</sup>.
2.  **Minimum Viable Product (MVP):** "노력 대비 효과" 최적의 조합인 **Hybrid Retrieval**과 **Re‑ranking**을 도입한다<sup>[\[5\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-weaviate-hybrid-5)[\[8\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-nogueira2019-8)</sup>.
3.  **Production:** 쿼리 변환(**HyDE**, **Multi‑Query**)을 추가하고, 필요 시 **Query Routing**을 적용한다. 가관측성(observability)(검색·재순위화·응답 로깅)과 A/B 테스트를 설정한다<sup>[\[3\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-langchain-multiquery-3)[\[10\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-llama-router-10)</sup>.

### 핵심 구성 요소

- **청킹(Chunking):** 품질에 가장 중요한 요소 중 하나다. 단순한 고정 크기 방식은 의미 단위를 자주 분리한다. 구조 지향적(마크업 기반) 또는 재귀적 분할기(문단 → 문장 → 단어)를 권장한다<sup>[\[17\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-rcsplit-17)[\[18\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-llama-hier-18)</sup>.
- **Embedding 및 메타데이터:** 각 청크에 document_id, 페이지/섹션, 제목, 날짜를 함께 저장한다. 이는 필터링과 정확한 출처 인용에 필수적이다.
- **하이브리드 검색 및 재순위화:** 퓨전(또는 RRF)을 활용한 BM25+벡터를 사용하고, 소규모 후보 풀에 대해 크로스 인코더로 재순위화를 수행한다<sup>[\[5\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-weaviate-hybrid-5)[\[6\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-qdrant-hybrid-6)[\[8\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-nogueira2019-8)</sup>.
- **컨텍스트 패키징:** 긴 코퍼스에는 *Map‑Reduce*, *Refine* 또는 *Tree‑of‑Chunks*를 선택한다<sup>[\[16\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-raptor-16)[\[18\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-llama-hier-18)</sup>.

### 일반적인 실수 (안티패턴)

- **벡터 검색만 사용**하고 BM25 없음 → 코드·ID·약어에서 검색 실패<sup>[\[5\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-weaviate-hybrid-5)[\[7\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-milvus-fulltext-7)</sup>.
- **청크 크기가 너무 크거나 작음** → 컨텍스트 손실 또는 embedding "희석"<sup>[\[17\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-rcsplit-17)</sup>.
- **프로덕션에서 재순위화 없음** → LLM이 노이즈가 많은 컨텍스트를 수신<sup>[\[8\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-nogueira2019-8)</sup>.
- **가관측성 및 출처 추적 없음** → 오류 원인 분석 불가능(RAG 평가 참조).

## 품질 평가 및 지표

평가는 검색기 수준(오프라인)과 엔드투엔드(생성) 수준에서 수행된다.

### 검색기 지표

- **Hit Rate, Recall@k, MRR** — 관련 문서의 커버리지와 순위 위치.
- **Context Precision & Recall** — 검색된 컨텍스트가 "노이즈" 없이 필요한 내용을 얼마나 잘 커버하는지 (RAGAS에 구현됨)<sup>[\[19\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-ragas-19)</sup>.

### 생성기 지표 (엔드투엔드)

- **Faithfulness / Groundedness** — 응답이 제공된 컨텍스트에 얼마나 부합하는지.
- **Answer Relevancy (답변 관련성)** — 원래 질문과의 일치 정도.

지표 자동화를 위해 오픈소스 프레임워크가 활용된다: **RAGAS**, **TruLens** (*RAG triad*: 컨텍스트 관련성, groundedness, 답변 관련성), **DeepEval**<sup>[\[20\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-trulens-20)[\[21\]](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_note-deepeval-21)</sup>.

## 같이 보기

- Retrieval-Augmented Generation (RAG)
- 벡터 데이터베이스
- Embedding
- AI 에이전트
- GraphRAG
- MM-RAG
- LLM 평가 및 benchmark

## 참고 문헌

- Lewis, P., Perez, E., et al. (2020). *Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks*. NeurIPS. arXiv:2005.11401.
- Fan, W., Ding, Y., et al. (2024). *A Survey on RAG Meeting LLMs: Towards Retrieval-Augmented Large Language Models*. KDD. DOI:10.1145/3637528.3671470; arXiv:2405.06211.
- Gao, L., Ma, X., Lin, J., Callan, J. (2023). *Precise Zero‑Shot Dense Retrieval without Relevance Labels (HyDE)*. ACL 2023. ACL Anthology; arXiv:2212.10496.
- Nogueira, R., Cho, K. (2019). *Passage Re‑ranking with BERT*. arXiv:1901.04085.
- Weaviate Docs. *Hybrid search (BM25+Vector)*. <a href="https://docs.weaviate.io/weaviate/concepts/search/hybrid-search" class="external autonumber" rel="nofollow">[1]</a>.
- Qdrant Docs. *Hybrid Queries*. <a href="https://qdrant.tech/documentation/concepts/hybrid-queries/" class="external autonumber" rel="nofollow">[2]</a>.
- Milvus Docs. *Full‑Text Search* / *Hybrid Search*. <a href="https://milvus.io/docs/full-text-search.md" class="external autonumber" rel="nofollow">[3]</a> / <a href="https://milvus.io/docs/hybrid_search_with_milvus.md" class="external autonumber" rel="nofollow">[4]</a>.
- LangChain Docs. *MultiQueryRetriever*. <a href="https://python.langchain.com/docs/how_to/MultiQueryRetriever/" class="external autonumber" rel="nofollow">[5]</a>.
- Cohere Docs. *Rerank — best practices*. <a href="https://docs.cohere.com/docs/reranking-best-practices" class="external autonumber" rel="nofollow">[6]</a>.
- LlamaIndex Docs. *Routing (query routers/selectors)*. <a href="https://docs.llamaindex.ai/en/stable/module_guides/querying/router/" class="external autonumber" rel="nofollow">[7]</a>.
- Yao, S., et al. (2023). *ReAct: Synergizing Reasoning and Acting in Language Models*. ICLR. arXiv:2210.03629.
- Nakano, R., et al. (2021). *WebGPT: Browser‑assisted question‑answering with human feedback*. arXiv:2112.09332.
- Microsoft Research Blog. *GraphRAG: Unlocking LLM discovery on narrative private data*. (2024). <a href="https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/" class="external autonumber" rel="nofollow">[8]</a>.
- Microsoft Research. *Project GraphRAG*. (2024). <a href="https://www.microsoft.com/en-us/research/project/graphrag/" class="external autonumber" rel="nofollow">[9]</a>.
- Yu, S., et al. (2024). *VisRAG: Vision‑based Retrieval‑augmented Generation on Multi‑modality Documents*. arXiv:2410.10594.
- Sarthi, P., et al. (2024). *RAPTOR: Recursive Abstractive Processing for Tree‑Organized Retrieval*. arXiv:2401.18059.
- Es, S., et al. (2024). *RAGAs: Automated Evaluation of Retrieval Augmented Generation*. EACL (Demo). <a href="https://aclanthology.org/2024.eacl-demo.16/" class="external autonumber" rel="nofollow">[10]</a>.
- TruLens Docs. *RAG Triad*. <a href="https://www.trulens.org/getting_started/core_concepts/rag_triad/" class="external autonumber" rel="nofollow">[11]</a>.
- DeepEval (GitHub). *The LLM Evaluation Framework*. <a href="https://github.com/confident-ai/deepeval" class="external autonumber" rel="nofollow">[12]</a>.
- LangChain Docs. *RecursiveCharacterTextSplitter*. <a href="https://python.langchain.com/docs/how_to/recursive_text_splitter/" class="external autonumber" rel="nofollow">[13]</a>.
- LlamaIndex Docs. *HierarchicalNodeParser*; *Response Synthesis (Tree/Refine)*. <a href="https://docs.llamaindex.ai/en/stable/api/llama_index.core.node_parser.HierarchicalNodeParser.html" class="external autonumber" rel="nofollow">[14]</a>; <a href="https://docs.llamaindex.ai/en/stable/examples/low_level/response_synthesis/" class="external autonumber" rel="nofollow">[15]</a>.

## 주석

1.  <span id="cite_note-lewis2020-1">↑ <sup>[1.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-lewis2020_1-0)</sup> <sup>[1.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-lewis2020_1-1)</sup> <sup>[1.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-lewis2020_1-2)</sup> <sup>[1.3](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-lewis2020_1-3)</sup> Lewis, P., Perez, E., et al. (2020). *Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks*. NeurIPS. arXiv:2005.11401.</span>
2.  <span id="cite_note-survey2024-2">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-survey2024_2-0) Fan, W., Ding, Y., et al. (2024). *A Survey on RAG Meeting LLMs: Towards Retrieval-Augmented Large Language Models*. KDD. DOI:10.1145/3637528.3671470; arXiv:2405.06211.</span>
3.  <span id="cite_note-langchain-multiquery-3">↑ <sup>[3.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-langchain-multiquery_3-0)</sup> <sup>[3.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-langchain-multiquery_3-1)</sup> <sup>[3.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-langchain-multiquery_3-2)</sup> LangChain Docs. *MultiQueryRetriever*. <a href="https://python.langchain.com/docs/how_to/MultiQueryRetriever/" class="external free" rel="nofollow">https://python.langchain.com/docs/how_to/MultiQueryRetriever/</a></span>
4.  <span id="cite_note-hyde-4">↑ <sup>[4.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-hyde_4-0)</sup> <sup>[4.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-hyde_4-1)</sup> Gao, L., Ma, X., Lin, J., Callan, J. (2023). *Precise Zero‑Shot Dense Retrieval without Relevance Labels*. ACL 2023. arXiv:2212.10496; ACL Anthology: 2023.acl‑long.99.</span>
5.  <span id="cite_note-weaviate-hybrid-5">↑ <sup>[5.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-weaviate-hybrid_5-0)</sup> <sup>[5.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-weaviate-hybrid_5-1)</sup> <sup>[5.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-weaviate-hybrid_5-2)</sup> <sup>[5.3](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-weaviate-hybrid_5-3)</sup> <sup>[5.4](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-weaviate-hybrid_5-4)</sup> Weaviate Docs. *Hybrid search (BM25+Vector)*. <a href="https://docs.weaviate.io/weaviate/concepts/search/hybrid-search" class="external free" rel="nofollow">https://docs.weaviate.io/weaviate/concepts/search/hybrid-search</a></span>
6.  <span id="cite_note-qdrant-hybrid-6">↑ <sup>[6.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-qdrant-hybrid_6-0)</sup> <sup>[6.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-qdrant-hybrid_6-1)</sup> <sup>[6.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-qdrant-hybrid_6-2)</sup> Qdrant Docs. *Hybrid Queries*. <a href="https://qdrant.tech/documentation/concepts/hybrid-queries/" class="external free" rel="nofollow">https://qdrant.tech/documentation/concepts/hybrid-queries/</a></span>
7.  <span id="cite_note-milvus-fulltext-7">↑ <sup>[7.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-milvus-fulltext_7-0)</sup> <sup>[7.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-milvus-fulltext_7-1)</sup> <sup>[7.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-milvus-fulltext_7-2)</sup> Milvus Docs. *Full‑Text Search* и *Hybrid Search*. <a href="https://milvus.io/docs/full-text-search.md" class="external free" rel="nofollow">https://milvus.io/docs/full-text-search.md</a>; <a href="https://milvus.io/docs/hybrid_search_with_milvus.md" class="external free" rel="nofollow">https://milvus.io/docs/hybrid_search_with_milvus.md</a></span>
8.  <span id="cite_note-nogueira2019-8">↑ <sup>[8.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-nogueira2019_8-0)</sup> <sup>[8.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-nogueira2019_8-1)</sup> <sup>[8.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-nogueira2019_8-2)</sup> <sup>[8.3](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-nogueira2019_8-3)</sup> <sup>[8.4](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-nogueira2019_8-4)</sup> Nogueira, R., Cho, K. (2019). *Passage Re‑ranking with BERT*. arXiv:1901.04085.</span>
9.  <span id="cite_note-cohere-rerank-9">↑ <sup>[9.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-cohere-rerank_9-0)</sup> <sup>[9.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-cohere-rerank_9-1)</sup> Cohere Docs. *Rerank — best practices*. <a href="https://docs.cohere.com/docs/reranking-best-practices" class="external free" rel="nofollow">https://docs.cohere.com/docs/reranking-best-practices</a></span>
10. <span id="cite_note-llama-router-10">↑ <sup>[10.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-llama-router_10-0)</sup> <sup>[10.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-llama-router_10-1)</sup> <sup>[10.2](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-llama-router_10-2)</sup> LlamaIndex Docs. *Routing (query routers/selectors)*. <a href="https://docs.llamaindex.ai/en/stable/module_guides/querying/router/" class="external free" rel="nofollow">https://docs.llamaindex.ai/en/stable/module_guides/querying/router/</a></span>
11. <span id="cite_note-react-11">↑ <sup>[11.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-react_11-0)</sup> <sup>[11.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-react_11-1)</sup> Yao, S., et al. (2023). *ReAct: Synergizing Reasoning and Acting in Language Models*. ICLR 2023. arXiv:2210.03629.</span>
12. <span id="cite_note-webgpt-12">↑ <sup>[12.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-webgpt_12-0)</sup> <sup>[12.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-webgpt_12-1)</sup> Nakano, R., et al. (2021). *WebGPT: Browser‑assisted question‑answering with human feedback*. arXiv:2112.09332.</span>
13. <span id="cite_note-graphrag-13">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-graphrag_13-0) Microsoft Research Blog. *GraphRAG: Unlocking LLM discovery on narrative private data*. 2024. <a href="https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/" class="external free" rel="nofollow">https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/</a></span>
14. <span id="cite_note-graphrag-project-14">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-graphrag-project_14-0) Microsoft Research. *Project GraphRAG*. <a href="https://www.microsoft.com/en-us/research/project/graphrag/" class="external free" rel="nofollow">https://www.microsoft.com/en-us/research/project/graphrag/</a></span>
15. <span id="cite_note-visrag-15">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-visrag_15-0) Yu, S., et al. (2024). *VisRAG: Vision‑based Retrieval‑augmented Generation on Multi‑modality Documents*. arXiv:2410.10594; OpenReview: zG459X3Xge.</span>
16. <span id="cite_note-raptor-16">↑ <sup>[16.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-raptor_16-0)</sup> <sup>[16.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-raptor_16-1)</sup> Sarthi, P., et al. (2024). *RAPTOR: Recursive Abstractive Processing for Tree‑Organized Retrieval*. arXiv:2401.18059.</span>
17. <span id="cite_note-rcsplit-17">↑ <sup>[17.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-rcsplit_17-0)</sup> <sup>[17.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-rcsplit_17-1)</sup> LangChain Docs. *RecursiveCharacterTextSplitter*. <a href="https://python.langchain.com/docs/how_to/recursive_text_splitter/" class="external free" rel="nofollow">https://python.langchain.com/docs/how_to/recursive_text_splitter/</a></span>
18. <span id="cite_note-llama-hier-18">↑ <sup>[18.0](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-llama-hier_18-0)</sup> <sup>[18.1](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-llama-hier_18-1)</sup> LlamaIndex Docs. *HierarchicalNodeParser* и *Tree Summarization*. <a href="https://docs.llamaindex.ai/en/stable/api/llama_index.core.node_parser.HierarchicalNodeParser.html" class="external free" rel="nofollow">https://docs.llamaindex.ai/en/stable/api/llama_index.core.node_parser.HierarchicalNodeParser.html</a>; <a href="https://docs.llamaindex.ai/en/stable/examples/low_level/response_synthesis/" class="external free" rel="nofollow">https://docs.llamaindex.ai/en/stable/examples/low_level/response_synthesis/</a></span>
19. <span id="cite_note-ragas-19">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-ragas_19-0) Es, S., et al. (2024). *RAGAs: Automated Evaluation of Retrieval Augmented Generation*. EACL (Demo). <a href="https://aclanthology.org/2024.eacl-demo.16/" class="external free" rel="nofollow">https://aclanthology.org/2024.eacl-demo.16/</a></span>
20. <span id="cite_note-trulens-20">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-trulens_20-0) TruLens Docs. *RAG Triad*. <a href="https://www.trulens.org/getting_started/core_concepts/rag_triad/" class="external free" rel="nofollow">https://www.trulens.org/getting_started/core_concepts/rag_triad/</a></span>
21. <span id="cite_note-deepeval-21">[↑](https://systems-analysis.info/int/RAG_patterns_(KO)#cite_ref-deepeval_21-0) DeepEval (GitHub). <a href="https://github.com/confident-ai/deepeval" class="external free" rel="nofollow">https://github.com/confident-ai/deepeval</a></span>
