---
title: "Hybrid retrieval (KO)"
source: "https://systems-analysis.info/int/Hybrid_retrieval_(KO)"
wiki: "systems-analysis.info/int"
article: "Hybrid_retrieval_(KO)"
language: "ko"
categories:
  - "Category:Korean"
  - "Category:Large language models"
  - "Category:Prompt engineering"
revision_id: 3091
wiki_created_at: 2026-09-06T23:15:34Z
wiki_modified_at: 2026-09-06T23:15:34Z
downloaded_at: 2026-09-07T22:55:01Z
---

# Hybrid retrieval (KO)

**Hybrid Retrieval (하이브리드 리트리벌)** — 검색 결과의 재현율과 정밀도를 높이기 위해 어휘적(sparse) 신호와 의미적(dense/late‑interaction) 신호를 결합하는 정보 검색 방법의 한 범주이다. 하이브리드 방식은 정확한 용어 일치(BM25/TF-IDF)와 벡터 유사도(bi-encoder, 다중 벡터 late-interaction 모델)의 장점을 결합하며, 스케일이 서로 다른 스코어링에 강건한 랭킹 융합 기법(예: Reciprocal Rank Fusion, CombSUM/CombMNZ)과 cross-encoder 재순위화를 활용한다.<sup>[\[1\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-1)[\[2\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-2)[\[3\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-3)</sup>

## 정의 및 동기

*하이브리드 리트리벌*이란 두 개(이상)의 독립적인 신호 채널에 대해 병렬 또는 캐스케이드 방식으로 검색을 수행한 뒤 결과를 융합하거나 재순위화하는 방법이다. 주요 동기는 다음과 같다: (i) 동의어·패러프레이즈로 인한 '용어 격차' 극복, (ii) 오탈자·형태소 변형에 대한 강건성 확보, (iii) 특정 코드·식별자 검색(sparse 모델이 강점을 보이는 영역), (iv) 새로운 도메인·언어로의 전이(dense 모델이 의미적 일반화를 제공하는 영역).<sup>[\[4\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-4)[\[5\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-5)[\[6\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-6)</sup>

## 하이브리드 검색의 구성 요소

### 어휘적(sparse)

- **고전 모델.** TF-IDF와 BM25/BM25F는 역색인 기반의 표준 기준선 방법이다. BM25는 확률적 PRF 프레임워크에 근거하며 1단계 랭킹에서 널리 사용된다.<sup>[\[7\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-7)</sup>
- **학습 가능한 sparse 모델.**
  - **SPLADE / SPLADE++/v3.** 희소성 정규화를 적용한 MLM 헤드를 통해 용어 확장 및 가중치를 학습하는 신경 sparse 모델로, 강력한 성능과 우수한 전이 가능성(BEIR)을 보인다.<sup>[\[8\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-8)[\[9\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-9)[\[10\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-10)</sup>
  - **uniCOIL/COIL.** 문맥화된 역색인 목록과 그 간소화 버전인 *uniCOIL*으로, 고전적인 역색인과 호환된다.<sup>[\[11\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-11)</sup>

### 의미적(dense/late‑interaction)

- **Bi‑encoder (단일 벡터).** 질의와 문서를 벡터 모델로 각각 인코딩하며 유사도는 dot-product/MIPS로 계산한다. 대표 예시: DPR,<sup>[\[12\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-12)</sup> ANCE,<sup>[\[13\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-13)</sup> Contriever,<sup>[\[14\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-14)</sup> GTR,<sup>[\[15\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-15)</sup> E5.<sup>[\[16\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-16)</sup>
- **Late‑interaction (다중 벡터).** '늦은' 상호작용 시점에 토큰 수준의 매칭을 모델링한다: ColBERT/ColBERTv2. 트레이드오프로 더 높은 정밀도를 얻는 대신 인덱스 크기와 레이턴시가 증가하며, 이는 엔지니어링 드라이버(PLAID, WARP)로 완화된다.<sup>[\[17\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-17)[\[18\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-18)[\[19\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-19)</sup>

## 하이브리드화 및 랭킹 융합 방식

- **병렬 검색 및 후보 목록 결합.** sparse와 dense 후보 목록을 각자의 내부 스코어와 함께 독립적으로 생성한 후 랭킹을 융합한다.<sup>[\[20\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-20)</sup>
- **RRF (Reciprocal Rank Fusion).** 서로 비교하기 어려운 스코어에 강건한 기법으로, 역순위를 합산한다:

${RRF}(d) = \sum\limits_{i = 1}^{m}\frac{1}{k + {rank}_{i}(d)}$, 일반적으로 $k \approx 60$.<sup>[\[21\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-21)</sup> Elasticsearch/OpenSearch 등 상용 엔진에서 내장 리트리버/프로세서로 지원된다.<sup>[\[22\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-22)[\[23\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-23)</sup>

- **CombSUM/CombMNZ 등.** 필요시 정규화를 적용하는 고전적인 '스코어 합산' 함수들이다.<sup>[\[24\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-24)[\[25\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-25)[\[26\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-26)</sup>
- **가중 선형 혼합.**

$S(d) = \alpha \cdot S_{\text{sparse}}(d) + (1 - \alpha) \cdot S_{\text{dense}}(d)$, $\alpha \in \lbrack 0,1\rbrack$. $\alpha$의 선택은 고정되거나 학습을 통해 결정(컬렉션별 또는 질의별)될 수 있다.<sup>[\[27\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-27)</sup>

- **스코어 정규화.** CombSUM/CombMNZ에서는 스케일 불일치 해소를 위해 min‑max, z‑score 등이 자주 사용되며,<sup>[\[28\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-28)</sup> RRF는 순위만을 기반으로 하는 대안적 접근이다.
- **동적/적응형 가중치.** 채널 선택 및 가중치 결정을 위한 질의 라우팅(query routing), 질의 특징, LTR 모델을 활용한다. 최근 연구들은 단순한 학습 혼합이 RRF를 능가하는 경우가 많으며 정규화에 둔감하다는 점을 보여 준다.<sup>[\[29\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-29)</sup>

## 재순위화 및 다단계 파이프라인

하이브리드 시스템은 일반적으로 *retrieval → fusion → rerank* 구조로 구축된다. 재순위화에는 다음이 활용된다:

- **Cross‑encoder (BERT/T5).** 가장 정확하지만 비용이 높다: MonoBERT/MonoT5를 이용한 상위 N개 후보 재순위화.<sup>[\[30\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-30)[\[31\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-31)</sup>
- **Late‑interaction을 재순위화기로 활용.** ColBERT 계열은 재순위화기로도 사용 가능하며, 최신 가속기(PLAID, WARP)가 품질 손실 없이 레이턴시를 감소시킨다.<sup>[\[32\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-32)[\[33\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-33)</sup>

*품질 ↔ 레이턴시/비용* 트레이드오프는 RAG와 엄격한 SLA 환경(p95/p99 꼬리 지연 참조)에서 특히 중요하다.<sup>[\[34\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-34)</sup>

## 벤치마크 평가

- **BEIR.** 리트리버의 zero-shot/out-of-domain 평가를 위한 다양한 컬렉션·과제로 구성된 통합 벤치마크(예: TREC‑COVID, NFCorpus, NQ, HotpotQA, FiQA‑2018, DBPedia‑entity, ArguAna, Webis‑Touché‑2020, FEVER/Climate‑FEVER, Scidocs, SciFact, CQADupStack 등).<sup>[\[35\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-35)</sup>
- **TREC Deep Learning / MS MARCO.** 대규모 데이터 환경에서 리트리버와 재순위화기를 학습·평가하기 위한 고전적 리소스.<sup>[\[36\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-36)[\[37\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-37)[\[38\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-38)</sup>
- **품질 지표.** nDCG@k, Recall@k, MRR; 성능 지표로는 latency p50/p95/p99, QPS; 운영 비용으로는 메모리/비용(CPU/GPU, 인덱스).<sup>[\[39\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-39)[\[40\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-40)</sup>
- **어블레이션.** 각 채널·가중치의 기여도, RRF의 $k$ 파라미터 및 혼합 시 $\alpha$에 대한 민감도를 고정하여 측정하고, 패러프레이즈와 OOD 이동에 대한 강건성을 평가할 것을 권장한다.<sup>[\[41\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-41)[\[42\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-42)</sup>

## 엔지니어링 측면 및 프로덕션 실천

- **인덱스 및 ANN.** MIPS/코사인 유사도를 위한 FAISS(Flat/HNSW/IVF‑PQ), HNSW, ScaNN.<sup>[\[43\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-43)[\[44\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-44)[\[45\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-45)</sup>
- **IR 스택.** sparse/dense 및 하이브리드 파이프라인을 위한 Lucene/Anserini/Pyserini; BEIR 기반 '원버튼' 재현성.<sup>[\[46\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-46)[\[47\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-47)</sup>
- **벡터 데이터베이스 및 검색 엔진.** Qdrant, Weaviate, pgvector/PostgreSQL, Vespa, Elasticsearch/OpenSearch는 하이브리드 검색 네이티브 모드(BM25F+vector) 및/또는 RRF/선형 혼합을 지원한다.<sup>[\[48\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-48)[\[49\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-49)[\[50\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-50)[\[51\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-51)[\[52\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-52)</sup>
- **RAG 패턴.** 아키텍처: **retrieval → fusion → rerank → LLM 컨텍스트** (토큰 제한 및 출처 추적 포함).<sup>[\[53\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-53)</sup>
- **인덱스 갱신, 중복 제거, 토크나이제이션.** BM25와 벡터화기 간 토크나이제이션을 일치시키는 것이 중요하며, 혼합 전 스코어 조정(정규화/스케일링)이 필요하다.<sup>[\[54\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-54)</sup>

## 한계 및 미해결 과제

- **전이 가능성 및 다국어 지원.** Dense 모델(GTR/E5)은 전이 성능을 향상시키지만 도메인·언어에 민감하며, sparse 모델(SPLADE)은 OOD 환경에서 더 강건한 경우가 많다.<sup>[\[55\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-55)[\[56\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-56)</sup>
- **LLM 통합 및 환각.** 하이브리드 리트리벌은 RAG 컨텍스트에서의 누락과 노이즈를 줄이지만 환각을 완전히 제거하지는 못하며, 엄격한 재순위화기와 출처 필터링이 필요하다.<sup>[\[57\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-57)</sup>
- **비용 및 개인정보 보호.** 다중 벡터 인덱스 저장, 압축, 암호화, on-prem 스택 및 TCO 평가.
- **트렌드.** 문서·질의 확장 기법으로서의 HyDE/doc2query/PRF;<sup>[\[58\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-58)[\[59\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-59)</sup> 혼합 학습(per-query $\alpha$), 더 효율적인 late-interaction(PLAID/WARP), 장문서 및 다중 벡터 인덱스.<sup>[\[60\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-60)[\[61\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-61)</sup>

## 방법 비교표

2025‑09‑10 기준 (BEIR *trec‑covid* 컬렉션 예시; nDCG@10 / Recall@100):<sup>[\[62\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-62)</sup>

| 방법                        | 유형 (sparse/dense/hybrid) | 아이디어/모델               | 융합 방식                 | 재순위화기           | nDCG@10 / R@100                                      | 레이턴시 (상대적) | 출처                                                                                                                                                                                                                                                  |
|-----------------------------|----------------------------|-----------------------------|---------------------------|----------------------|------------------------------------------------------|-------------------|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| BM25                        | sparse                     | 정확한 용어 일치 (PRF/BM25) | —                         | —                    | 0.595 / 0.109                                        | 매우 낮음         | <sup>[\[63\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-63)[\[64\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-64)</sup>                                                                               |
| SPLADE++ (ED)               | sparse (learned)           | 희소 용어 확장/가중치       | —                         | —                    | 0.727 / 0.128                                        | 낮음–중간         | <sup>[\[65\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-65)[\[66\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-66)</sup>                                                                               |
| Contriever (MS MARCO FT)    | dense                      | 대조 학습 bi-encoder        | —                         | —                    | 0.596 / 0.091                                        | 중간              | <sup>[\[67\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-67)[\[68\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-68)</sup>                                                                               |
| BGE‑base‑en‑v1.5            | dense                      | 강력한 범용 임베더          | —                         | —                    | 0.781 / 0.141                                        | 중간              | <sup>[\[69\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-69)</sup>                                                                                                                                                             |
| Cohere embed‑english‑v3.0   | dense                      | 상용 텍스트 embedding 모델  | —                         | —                    | 0.818 / 0.159                                        | 중간              | <sup>[\[70\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-70)</sup>                                                                                                                                                             |
| BM25 + dense (예: BM25+BGE) | hybrid                     | 병렬 리트리벌 + 목록 융합   | RRF (k≈60) 또는 가중 혼합 | 선택: MonoT5/ColBERT | (구현에 따라 다름; 통상적으로 최선의 단일 채널 초과) | 중간              | <sup>[\[71\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-71)[\[72\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-72)[\[73\]](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_note-73)</sup> |

*trec‑covid*에서의 방법 비교

비고: 마지막 행은 방식을 예시하며, 정확한 수치는 임베더 선택, 정규화, 융합 파라미터에 따라 달라진다(출처 및 Pyserini 재현 가능 스크립트 참조).

## 참고 문헌

- Manning, C.D., Raghavan, P., Schütze, H. (2008). *Introduction to Information Retrieval*. Cambridge University Press. ISBN 978‑0521865715.
- Robertson, S., Zaragoza, H. (2009). *The Probabilistic Relevance Framework: BM25 and Beyond*. Foundations and Trends in Information Retrieval 3(4):333–389. DOI:10.1561/1500000019.
- Lin, J. et al. (2021). *Pyserini: A Python Toolkit for Reproducible IR*. SIGIR.
- Järvelин, K., Kekäläinen, J. (2002). *Cumulated Gain‑Based Evaluation of IR Techniques*. Information Retrieval 6:241–256. DOI:10.1023/A:1016043826386.
- Dean, J., Barroso, L.A. (2013). *The Tail at Scale*. CACM 56(2):74–80. DOI:10.1145/2408776.2408794.

## 외부 링크

- Pyserini / Anserini: github.com/castorini/pyserini • github.com/castorini/anserini
- FAISS: arXiv:1702.08734
- Weaviate (Hybrid search): docs.weaviate.io/weaviate/search/hybrid
- pgvector: github.com/pgvector/pgvector
- Vespa (Hybrid search tutorial): docs.vespa.ai/en/tutorials/hybrid-search.html

## 주석

1.  <span id="cite_note-1">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-1) Robertson, S., Zaragoza, H. (2009). *The Probabilistic Relevance Framework: BM25 and Beyond*. Foundations and Trends in Information Retrieval, 3(4), 333–389. DOI:10.1561/1500000019.</span>
2.  <span id="cite_note-2">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-2) Cormack, G.V., Clarke, C.L.A., Büttcher, S. (2009). *Reciprocal Rank Fusion Outperforms Condorcet and Individual Rank Learning Methods*. SIGIR 2009, 758–759. <a href="https://cormack.uwaterloo.ca/cormacksigir09-rrf.pdf" class="external text" rel="nofollow">PDF</a>.</span>
3.  <span id="cite_note-3">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-3) Bruch, S., Gai, S., Ingber, A. (2023). *An Analysis of Fusion Functions for Hybrid Retrieval*. ACM TOIS 42(1):1–35. DOI:10.1145/3596512 • arXiv:2210.11934.</span>
4.  <span id="cite_note-4">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-4) Manning, C.D., Raghavan, P., Schütze, H. (2008). *Introduction to Information Retrieval*. Cambridge University Press. ISBN 978‑0521865715 (см. главы о TF‑IDF, оценке и проблеме *vocabulary mismatch*).</span>
5.  <span id="cite_note-5">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-5) Izacard, G. et al. (2022). *Unsupervised Dense Information Retrieval with Contrastive Learning (Contriever)*. TACL 10:1089–1108. arXiv:2112.09118.</span>
6.  <span id="cite_note-6">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-6) Wang, L. et al. (2022/2024). *Text Embeddings by Weakly‑Supervised Contrastive Pre‑training (E5)*. arXiv:2212.03533.</span>
7.  <span id="cite_note-7">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-7) Robertson, S., Zaragoza, H. (2009). *The Probabilistic Relevance Framework: BM25 and Beyond*. DOI:10.1561/1500000019.</span>
8.  <span id="cite_note-8">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-8) Formal, T., Piwowarski, B., Clinchant, S. (2021). *SPLADE: Sparse Lexical and Expansion Model for First Stage Ranking*. arXiv:2107.05720.</span>
9.  <span id="cite_note-9">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-9) Formal, T. et al. (2022). *Making Sparse Neural IR Models More Effective*. Findings of EMNLP. arXiv:2205.04733.</span>
10. <span id="cite_note-10">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-10) Formal, T. et al. (2024). *SPLADE‑v3: New baselines for SPLADE*. arXiv:2403.06789.</span>
11. <span id="cite_note-11">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-11) Lin, J., Ma, X. (2021). *A Few Brief Notes on DeepImpact, COIL, and uniCOIL*. arXiv:2106.14807.</span>
12. <span id="cite_note-12">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-12) Karpukhin, V. et al. (2020). *Dense Passage Retrieval for Open‑Domain QA*. EMNLP. arXiv:2004.04906.</span>
13. <span id="cite_note-13">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-13) Xiong, L. et al. (2021). *Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval (ANCE)*. ICLR. arXiv:2007.00808.</span>
14. <span id="cite_note-14">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-14) Izacard, G. et al. (2022). TACL. arXiv:2112.09118.</span>
15. <span id="cite_note-15">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-15) Ni, J. et al. (2021/2022). *Large Dual Encoders Are Generalizable Retrievers (GTR)*. EMNLP. arXiv:2112.07899.</span>
16. <span id="cite_note-16">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-16) Wang, L. et al. (2022/2024). arXiv:2212.03533.</span>
17. <span id="cite_note-17">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-17) Khattab, O., Zaharia, M. (2020). *ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT*. SIGIR. arXiv:2004.12832.</span>
18. <span id="cite_note-18">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-18) Santhanam, K. et al. (2022). *ColBERTv2 & PLAID*. NAACL/ArXiv. arXiv:2112.01488; arXiv:2205.09707.</span>
19. <span id="cite_note-19">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-19) Scheerer, J.L. et al. (2025). *WARP: An Efficient Engine for Multi‑Vector Retrieval*. arXiv:2501.17788.</span>
20. <span id="cite_note-20">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-20) Lin, J. et al. (2021). *Pyserini: A Python Toolkit for Reproducible IR with Sparse and Dense Representations*. SIGIR. <a href="https://cs.uwaterloo.ca/~jimmylin/publications/Lin_etal_SIGIR2021_Pyserini.pdf" class="external text" rel="nofollow">PDF</a>.</span>
21. <span id="cite_note-21">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-21) Cormack, G.V., Clarke, C.L.A., Büttcher, S. (2009). SIGIR. <a href="https://cormack.uwaterloo.ca/cormacksigir09-rrf.pdf" class="external text" rel="nofollow">PDF</a>.</span>
22. <span id="cite_note-22">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-22) Elastic Docs. *Reciprocal Rank Fusion*. (доступ 2025‑09‑10). <a href="https://www.elastic.co/docs/reference/elasticsearch/rest-apis/reciprocal-rank-fusion" class="external text" rel="nofollow">elastic.co/docs/.../reciprocal-rank-fusion</a>.</span>
23. <span id="cite_note-23">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-23) OpenSearch Docs. *Score ranker processor (RRF)*. (доступ 2025‑09‑10). <a href="https://docs.opensearch.org/3.0/search-plugins/search-pipelines/score-ranker-processor/" class="external text" rel="nofollow">docs.opensearch.org/.../score-ranker-processor/</a>.</span>
24. <span id="cite_note-24">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-24) Fox, E.A., Shaw, J.A. (1994). *Combination of Multiple Searches*. TREC‑2, NIST SP 500‑215, 243–252. <a href="https://www.khoury.northeastern.edu/home/jaa/CSG339.06F/resources/fox94combination.pdf" class="external text" rel="nofollow">PDF</a>.</span>
25. <span id="cite_note-25">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-25) Lee, J.H. (1997). *Analyses of Multiple Evidence Combination*. SIGIR, 267–276. DOI:10.1145/258525.258587.</span>
26. <span id="cite_note-26">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-26) Hsu, D.F., Taksa, I. (2005). *Comparing Rank and Score Combination Methods for Data Fusion in IR*. (Tech. report). <a href="https://ccc.inaoep.mx/~villasen/bib/Hsu-FusionInIR07.pdf" class="external text" rel="nofollow">PDF</a>.</span>
27. <span id="cite_note-27">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-27) Bruch, S., Gai, S., Ingber, A. (2023). TOIS. DOI:10.1145/3596512.</span>
28. <span id="cite_note-28">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-28) Hsu, D.F., Taksa, I. (2005). см. выше.</span>
29. <span id="cite_note-29">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-29) Bruch, S., Gai, S., Ingber, A. (2023). TOIS. DOI:10.1145/3596512.</span>
30. <span id="cite_note-30">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-30) Nogueira, R., Cho, K. (2019). *Passage Re‑ranking with BERT*. arXiv:1901.04085.</span>
31. <span id="cite_note-31">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-31) Nogueira, R., Jiang, Z., Lin, J. (2020). *Document Ranking with a Pretrained Sequence‑to‑Sequence Model (MonoT5)*. Findings of EMNLP. arXiv:2003.06713.</span>
32. <span id="cite_note-32">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-32) Santhanam, K. et al. (2022). arXiv:2205.09707.</span>
33. <span id="cite_note-33">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-33) Scheerer, J.L. et al. (2025). arXiv:2501.17788.</span>
34. <span id="cite_note-34">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-34) Dean, J., Barroso, L.A. (2013). *The Tail at Scale*. CACM 56(2):74–80. DOI:10.1145/2408776.2408794.</span>
35. <span id="cite_note-35">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-35) Thakur, N. et al. (2021). *BEIR: A Heterogeneous Benchmark for Zero‑shot Evaluation of IR Models*. NeurIPS Datasets & Benchmarks. arXiv:2104.08663.</span>
36. <span id="cite_note-36">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-36) Craswell, N. et al. (2020). *Overview of the TREC 2019 Deep Learning Track*. arXiv:2003.07820.</span>
37. <span id="cite_note-37">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-37) Craswell, N. et al. (2021). *Overview of the TREC 2020 Deep Learning Track*. arXiv:2102.07662.</span>
38. <span id="cite_note-38">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-38) Bajaj, P. et al. (2016). *MS MARCO: A Human Generated MAchine Reading COmprehension Dataset*. arXiv:1611.09268.</span>
39. <span id="cite_note-39">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-39) Järvelin, K., Kekäläinen, J. (2002). *Cumulated Gain‑Based Evaluation of IR Techniques*. Information Retrieval 6:241–256. DOI:10.1023/A:1016043826386.</span>
40. <span id="cite_note-40">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-40) Dean, J., Barroso, L.A. (2013). CACM. DOI:10.1145/2408776.2408794.</span>
41. <span id="cite_note-41">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-41) Bruch, S. et al. (2023). DOI:10.1145/3596512.</span>
42. <span id="cite_note-42">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-42) Ni, J. et al. (2021/2022). arXiv:2112.07899.</span>
43. <span id="cite_note-43">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-43) Johnson, J., Douze, M., Jégou, H. (2017). *Billion‑scale Similarity Search with GPUs (FAISS)*. arXiv:1702.08734.</span>
44. <span id="cite_note-44">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-44) Malkov, Y., Yashunin, D. (2020). *HNSW*. IEEE TPAMI 42(4):824–836. DOI:10.1109/TPAMI.2018.2889473.</span>
45. <span id="cite_note-45">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-45) Guo, R. et al. (2020). *ScaNN: Efficient Vector Similarity Search at Scale*. arXiv:1908.10396.</span>
46. <span id="cite_note-46">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-46) Yang, P., Fang, H., Lin, J. (2018). *Anserini: Reproducible IR Research with Lucene*. JDIQ 10(4):1–20. DOI:10.1145/3239571.</span>
47. <span id="cite_note-47">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-47) Lin, J. et al. (2021). SIGIR. <a href="https://cs.uwaterloo.ca/~jimmylin/publications/Lin_etal_SIGIR2021_Pyserini.pdf" class="external text" rel="nofollow">PDF</a>.</span>
48. <span id="cite_note-48">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-48) Qdrant Docs. *Hybrid queries (RRF, DBSF)*. (доступ 2025‑09‑10). <a href="https://qdrant.tech/documentation/concepts/hybrid-queries/" class="external text" rel="nofollow">qdrant.tech/.../hybrid-queries/</a>.</span>
49. <span id="cite_note-49">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-49) Weaviate Docs. *Hybrid search*. (доступ 2025‑09‑10). <a href="https://docs.weaviate.io/weaviate/search/hybrid" class="external text" rel="nofollow">docs.weaviate.io/weaviate/search/hybrid</a>.</span>
50. <span id="cite_note-50">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-50) pgvector GitHub. (доступ 2025‑09‑10). <a href="https://github.com/pgvector/pgvector" class="external text" rel="nofollow">github.com/pgvector/pgvector</a>.</span>
51. <span id="cite_note-51">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-51) Vespa Docs. *Hybrid Text Search Tutorial*. (доступ 2025‑09‑10). <a href="https://docs.vespa.ai/en/tutorials/hybrid-search.html" class="external text" rel="nofollow">docs.vespa.ai/.../hybrid-search.html</a>.</span>
52. <span id="cite_note-52">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-52) Elastic Docs. *Reciprocal Rank Fusion*. (доступ 2025‑09‑10). <a href="https://www.elastic.co/docs/reference/elasticsearch/rest-apis/reciprocal-rank-fusion" class="external text" rel="nofollow">elastic.co/docs/.../rrf</a>.</span>
53. <span id="cite_note-53">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-53) Lewis, P. et al. (2020). *Retrieval‑Augmented Generation for Knowledge‑Intensive NLP Tasks*. NeurIPS. arXiv:2005.11401.</span>
54. <span id="cite_note-54">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-54) Hsu, D.F., Taksa, I. (2005). см. выше.</span>
55. <span id="cite_note-55">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-55) Ni, J. et al. (2021/2022). arXiv:2112.07899.</span>
56. <span id="cite_note-56">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-56) Formal, T. et al. (2021, 2022, 2024). arXiv:2107.05720; 2205.04733; 2403.06789.</span>
57. <span id="cite_note-57">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-57) Lewis, P. et al. (2020). arXiv:2005.11401.</span>
58. <span id="cite_note-58">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-58) Gao, L. et al. (2023). *Precise Zero‑Shot Dense Retrieval without Relevance Labels (HyDE)*. ACL. arXiv:2212.10496.</span>
59. <span id="cite_note-59">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-59) Nogueira, R. et al. (2019). *Document Expansion by Query Prediction*. arXiv:1904.08375; *docTTTTTquery*. <a href="https://cs.uwaterloo.ca/~jimmylin/publications/Nogueira_Lin_2019_docTTTTTquery-v2.pdf" class="external text" rel="nofollow">PDF</a>.</span>
60. <span id="cite_note-60">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-60) Santhanam, K. et al. (2022). arXiv:2205.09707.</span>
61. <span id="cite_note-61">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-61) Scheerer, J.L. et al. (2025). arXiv:2501.17788.</span>
62. <span id="cite_note-62">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-62) Pyserini *BEIR Regressions* (доступ 2025‑09‑10): результаты по *trec‑covid* для BM25/SPLADE/Contriever/BGE/Cohere. <a href="https://castorini.github.io/pyserini/2cr/beir.html" class="external text" rel="nofollow">castorini.github.io/pyserini/2cr/beir.html</a>.</span>
63. <span id="cite_note-63">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-63) Robertson, S., Zaragoza, H. (2009). DOI:10.1561/1500000019.</span>
64. <span id="cite_note-64">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-64) Pyserini BEIR. См. ссылку выше.</span>
65. <span id="cite_note-65">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-65) Formal, T. et al. (2021, 2022). arXiv:2107.05720; 2205.04733.</span>
66. <span id="cite_note-66">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-66) Pyserini BEIR.</span>
67. <span id="cite_note-67">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-67) Izacard, G. et al. (2022). arXiv:2112.09118.</span>
68. <span id="cite_note-68">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-68) Pyserini BEIR.</span>
69. <span id="cite_note-69">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-69) Pyserini BEIR.</span>
70. <span id="cite_note-70">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-70) Pyserini BEIR.</span>
71. <span id="cite_note-71">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-71) Cormack et al. (2009). SIGIR. RRF.</span>
72. <span id="cite_note-72">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-72) Bruch et al. (2023). TOIS.</span>
73. <span id="cite_note-73">[↑](https://systems-analysis.info/int/Hybrid_retrieval_(KO)#cite_ref-73) Elastic/OpenSearch RRF Docs.</span>