---
title: "Hypothetical Document Embeddings (HyDE) (TR)"
source: "https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)"
wiki: "systems-analysis.info/int"
article: "Hypothetical_Document_Embeddings_(HyDE)_(TR)"
language: "tr"
categories:
  - "Category:Large language models"
  - "Category:Prompt engineering"
  - "Category:Turkish"
revision_id: 3125
wiki_created_at: 2026-09-06T23:16:05Z
wiki_modified_at: 2026-09-06T23:16:05Z
downloaded_at: 2026-09-07T22:55:20Z
---

# Hypothetical Document Embeddings (HyDE) (TR)

**Hypothetical Document Expansion (HyDE)** — vektörel retrieval ve retrieval‑augmented generation (RAG) yöntemini iyileştirmeye yönelik bir yöntemdir; bu yöntemde büyük dil modeli (LLM), özgün sorguya dayanarak bir "hipotetik belge" üretir; ardından bu metin bir encoder tarafından vektörleştirilir ve arama, elde edilen vektöre yakınlık temelinde gerçek belgeler arasında gerçekleştirilir. Bu yaklaşım, LLM tarafından kodlanan "ilgililik örüntülerini" kullanmayı ve bunları yoğun embedding'ler aracılığıyla bir derlem üzerinde "temellendirmeyi" mümkün kılar<sup>[\[1\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-1)</sup>.

## Tanım ve Sezgi

HyDE, arama görevini iki aşamaya ayırır:

\(1\) LLM, sorguya yönelik bir "ilgili yanıt örneği" (*hypothetical document*) oluşturarak ilgililik özelliklerini modeller;

\(2\) kontrastif bir encoder (örn. Contriever), bu metni bir vektöre dönüştürür; bu vektör aracılığıyla indeksten gerçek belgeler elde edilir. Üretilen metin olgusal hatalar içerebilir; ancak önemli olan, encoder tarafından yakalanan tematik ve terminolojik örüntülerdir<sup>[\[2\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-2)</sup>.

## Tarihçe ve Kaynaklar

Aramayı sentetik metinlerle genişletme fikri, sorgu genişletme ve sözde-ilgili geri bildirim (PRF) üzerine yapılan çalışmalara dayanır: Rocchio algoritması ve ilgililik dil modelleri<sup>[\[3\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-3)[\[4\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-4)</sup>. Yoğun retrieval için kontrastif eğitimli encoder'lar (Contriever)<sup>[\[5\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-5)</sup> ve Dense Passage Retrieval (DPR)<sup>[\[6\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-6)</sup> kullanılmıştır. BEIR benchmark'ı, zero‑shot değerlendirmeyi standart hale getirmiştir<sup>[\[7\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-7)</sup>. Bu bağlamda HyDE, encoder'ı yeniden eğitmeye gerek kalmaksızın LLM aracılığıyla sıfır atış moduna ilgililik bilgisini "taşımanın" bir yolu olarak önerilmiştir<sup>[\[8\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-8)</sup>.

## Yöntem ve Biçimselleştirme

Belge derleminin $\mathcal{D} = \{ d_{1},\ldots,d_{N}\}$ olduğunu ve bir metin encoder'ının $E:\text{text} \rightarrow {\mathbb{R}}^{n}$, belgeler için vektörel gösterimler $\mathbf{v}_{d} = E(d)$ tanımladığını varsayalım. Yakınlık ölçümü için kosinüs benzerliği ya da skaler çarpım kullanılır; önemli bir not: \*\*skaler çarpım, yalnızca her iki vektörün L2‑normu bire eşit olduğunda kosinüs benzerliğiyle örtüşür\*\* ($\|\mathbf{u}\| = \|\mathbf{v}\| = 1$)<sup>[\[9\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-9)</sup>.

HyDE, sorgu gösterimini LLM tarafından üretilen bir "hipotetik belge" aracılığıyla yeniden tanımlar. Biçimsel olarak:

$$
\begin{matrix}
 & \text{(1) Генерация гипотетического текста:} & & {\overset{\sim}{d}\; = \; G\!\left( q;\,{inst} \right),} \\
 & \text{(2) Эмбеддинг гипотетического текста:} & & {\mathbf{v}_{h}\; = \; E(\overset{\sim}{d}),} \\
 & \text{(3) Поиск ближайших соседей:} & & {\mathcal{R}_{k}(q)\; = \;{TopK}_{\, d \in \mathcal{D}}\; S\!\left( \mathbf{v}_{h},\mathbf{v}_{d} \right),}
\end{matrix}
$$

burada $G$ — $inst$ talimatına sahip LLM (örneğin: "Şu soruyu yanıtlayan bir paragraf yaz…"), $S$ — benzerlik ölçüsü (normalizasyonlu kosinüs veya IP), $\mathcal{R}_{k}(q)$ ise maksimum benzerliğe sahip $k$ belgeden oluşan kümedir<sup>[\[10\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-10)[\[11\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-11)</sup>.

Mühendislik uygulamalarında çoğunlukla \*\*birden fazla\*\* hipotetik metin üretilir ve bunların gösterimleri toplanır; bu da kararlılığı artırır:

$$
{\overset{\sim}{d}}^{(j)} = G\!\left( q;\,{inst},\xi_{j} \right),\quad\mathbf{v}_{h}\; = \;\frac{1}{m}\sum\limits_{j = 1}^{m}E\!\left( {\overset{\sim}{d}}^{(j)} \right),
$$

burada $\xi_{j}$ — stokastik kod çözme parametreleridir (örn. temperature/top‑p). Bu tür ensemble yaklaşımı, orta düzeyde gecikme artışıyla Recall'ı iyileştirir<sup>[\[12\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-12)</sup>.

### HyDE Temel Pipeline'ı

    # 1) prompt(query) -> hypothetical_doc
    # 2) embed(hypothetical_doc) -> v_h
    # 3) retrieve(index, v_h, k) -> candidates
    # 4) (optional) rerank(query, candidates) -> topN
    # 5) (для RAG) stuff / map-reduce / refine на topN

### Diğer Yöntemlerle İlişki (QE, doc2query, PRF)

- **QE (sorgu genişletme)** sorguya terimler ekler; HyDE bunun yerine yoğun encoder'larla daha uyumlu olan tüm bir "yarı-belge" üretir<sup>[\[13\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-13)</sup>.
- **doc2query / docTTTTTquery** indeksleme öncesinde sentetik sorgularla **belgeleri** genişletir<sup>[\[14\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-14)[\[15\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-15)</sup>; HyDE **sorguyu** anında genişletir ve yeniden indeksleme gerektirmez.
- **PRF** (Rocchio, Relevance LM), sorgu vektörünü en iyi sonuçlara göre günceller; HyDE "ilgililik örüntüsünü" doğrudan LLM'den çıkarır ve ardından derlem üzerinde retrieval aracılığıyla "temellendirir"<sup>[\[16\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-16)</sup>.

## RAG ve Yeniden Sıralamaya Entegrasyon

RAG'da HyDE, retrieval'ın ilk aşaması olarak uygulanır: hipotetik belge → embedding → k aday. Ardından yeniden sıralama uygulanır: BERT sınıfı cross-encoder'lar<sup>[\[17\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-17)</sup> veya geç etkileşimli ColBERT<sup>[\[18\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-18)</sup>. Liste birleştirme için (örn. hibrit BM25+vector) tipik olarak RRF (*reciprocal rank fusion*) kullanılır: $\operatorname{RRF}(d) = \sum\limits_{r \in \mathcal{R}}\frac{1}{k + \operatorname{rank}_{r}(d)},\qquad k \approx 60.$ RRF yöntemi, birleştirilmiş sıralamaların genel kalitesini tutarlı biçimde artırır<sup>[\[19\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-19)</sup>.

## Benchmark Değerlendirmesi (BEIR ve Diğerleri)

Özgün çalışma, HyDE'yi TREC DL'19/20 (web araması) ve BEIR koleksiyonlarının bir alt kümesinde (Scifact, ArguAna, TREC‑COVID, FiQA, DBPedia, TREC‑NEWS, Climate‑FEVER) sıfır atış modunda değerlendirmektedir. Sonuçların bir kısmı — *2023‑07 itibarıyla*:

| Yöntem                    | DL19                   | DL20                   | Kaynak                                                                                                           |
|---------------------------|------------------------|------------------------|------------------------------------------------------------------------------------------------------------------|
| BM25                      | 30.1 / 50.6 / 75.0     | 28.6 / 48.0 / 78.6     | <sup>[\[20\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-20)</sup> |
| Contriever (unsup.)       | 24.0 / 44.5 / 74.6     | 24.0 / 42.1 / 75.4     | <sup>[\[21\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-21)</sup> |
| **HyDE** (Contriever+LLM) | **41.8 / 61.3 / 88.0** | **38.2 / 57.9 / 84.4** | <sup>[\[22\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-22)</sup> |
| DPR (ft)                  | 36.5 / 62.2 / 76.9     | 41.8 / 65.3 / 81.4     | <sup>[\[23\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-23)</sup> |
| ANCE (ft)                 | 37.1 / 64.5 / 75.5     | 40.8 / 64.6 / 77.6     | <sup>[\[24\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-24)</sup> |

*TREC DL19/20 (web araması)* — mAP / nDCG@10 / Recall@1k

| Yöntem     | Scifact         | ArguAna         | TREC‑COVID      | FiQA        | DBPedia     | TREC‑NEWS       | Climate‑FEVER   | Kaynak                                                                                                           |
|------------|-----------------|-----------------|-----------------|-------------|-------------|-----------------|-----------------|------------------------------------------------------------------------------------------------------------------|
| BM25       | 67.9 / 92.5     | 39.7 / 93.2     | **59.5 / 49.8** | 23.6 / 54.0 | 31.8 / 46.8 | 39.5 / 44.7     | 16.5 / 42.5     | <sup>[\[25\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-25)</sup> |
| Contriever | 64.9 / 92.6     | 37.9 / 90.1     | 27.3 / 17.2     | 24.5 / 56.2 | 29.2 / 45.3 | 34.8 / 42.3     | 15.5 / 44.1     | <sup>[\[26\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-26)</sup> |
| **HyDE**   | **69.1 / 96.4** | **46.6 / 97.9** | 59.3 / 41.4     | 27.3 / 62.1 | 36.8 / 47.2 | **44.0 / 50.9** | **22.3 / 53.0** | <sup>[\[27\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-27)</sup> |

*BEIR (veri kümesi seçkisi)* — nDCG@10 / Recall@100

HyDE ayrıca Mr.TyDi çok dilli veri kümelerinde (sw/ko/ja/bn) mContriever'a kıyasla MRR@100 değerini de iyileştirmektedir<sup>[\[28\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-28)</sup>.

## Pratik Öneriler

HyDE'nin ne zaman kullanılacağı

- Sıfır atış/aktarım modları (ilgililik etiketi yok; alan "benzemezliği" eğitim derlemlerine kıyasla)<sup>[\[29\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-29)</sup>.
- Kabul edilebilir bir hassasiyet düzeyinde Recall@k artışı gerektiğinde — HyDE çoğunlukla vektör uzayındaki ilgili bölgeleri "açar"<sup>[\[30\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-30)</sup>.

Tipik Ayarlar

- **LLM ve prompt**: "Şu soruyu yanıtlayan bir paragraf yaz…" talimatı; orta düzey stokastisite (örn. *temperature*≈0.7)<sup>[\[31\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-31)</sup>.
- **Hipotetik metin sayısı**: 1–5; embedding'lerin ortalaması alınması kararlılığı artırır<sup>[\[32\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-32)</sup>.
- **Embedding modeli**: Fine-tuning olmaksızın (m)Contriever; fine-tuning uygulanmış encoder'lar da kullanılabilir (HyDE etkisi korunur)<sup>[\[33\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-33)</sup>.
- **Embedding normalizasyonu**: L2‑norm; iç çarpım kosinüs ile eşdeğerdir<sup>[\[34\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-34)</sup>.
- **Hibrit retrieval**: ardından yeniden sıralama ile BM25+vector<sup>[\[35\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-35)</sup>.
- **Yeniden sıralayıcı**: Cross-Encoder (BERT re‑ranker)<sup>[\[36\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-36)</sup> veya ColBERT<sup>[\[37\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-37)</sup>.
- **Farklı stratejilerin sonuçlarını birleştirme**: RRF (*k*≈60)<sup>[\[38\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-38)</sup>.

Kalite/Maliyet İzleme

- Retrieval: nDCG@k, Recall@k, MRR; uçtan uca RAG: EM/F1 veya *groundedness* metrikleri (RAGAS/TruLens)<sup>[\[39\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-39)[\[40\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-40)</sup>.
- Maliyet/gecikme: LLM üretimi ve (varsa) yeniden sıralama baskın unsurdur; "hipotetik" metin sayısı ve yanıt uzunluğuyla optimize edilir<sup>[\[41\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-41)</sup>.

## Sınırlamalar ve Açık Sorular

- **Hipotetik metnin halüsinasyonları**: LLM olgusal hatalar üretebilir; encoder ve derlem aracılığıyla "temellendirme" riski azaltır ancak tamamen ortadan kaldırmaz<sup>[\[42\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-42)</sup>.
- **Alan ve dil kısıtlamaları**: HyDE'nin sağladığı kazanım, son derece uzmanlaşmış alanlarda ve düşük kaynaklı dillerde azalmaktadır<sup>[\[43\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-43)</sup>.
- **Gecikme ve maliyet**: LLM üretimi ek gecikme ve token maliyeti getirir; bu durum çevrimiçi senaryolar ve uzun "hipotetik" metinler için kritik öneme sahiptir<sup>[\[44\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-44)</sup>.
- **Etik ve önyargılar**: Güvenli LLM'lerin ve filtrelemenin kullanılması tercih edilir<sup>[\[45\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-45)</sup>.

## Yöntemlerin Karşılaştırma Tablosu

| Yöntem                    | Sınıf                    | Metnin üretildiği yer                    | Encoder/İndeks        | Yeniden sıralayıcı (2. aşama) | Tipik metrikler (örnek)                            | Maliyet/Gecikme                                   | Kaynaklar                                                                                                                                                                                                             |
|---------------------------|--------------------------|------------------------------------------|-----------------------|-------------------------------|----------------------------------------------------|---------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| **HyDE**                  | Query→*hypo‑doc*         | Sorgu tarafında (LLM → paragraf)         | (m)Contriever; ANN    | BERT re‑rank / ColBERT / RRF  | DL19 nDCG@10≈61.3; DL20≈57.9; ArguAna nDCG@10≈46.6 | \+ LLM üretimi; + yeniden sıralama (isteğe bağlı) | <sup>[\[46\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-46)</sup>                                                                                                      |
| BM25                      | Sözcüksel                | —                                        | Ters çevrilmiş indeks | İsteğe bağlı                  | bkz. tablo (yukarıda)                              | Düşük (lexical)                                   | <sup>[\[47\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-47)</sup>                                                                                                      |
| DPR / ANCE                | Yoğun (ft)               | —                                        | Bi‑encoder; ANN       | İsteğe bağlı                  | DL19 nDCG@10≈62–65                                 | Orta (LLM yok)                                    | <sup>[\[48\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-48)[\[49\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-49)</sup> |
| doc2query / docTTTTTquery | Belge genişletme         | Koleksiyon tarafında (indeksleme öncesi) | BM25/sparse+expanded  | İsteğe bağlı                  | MS MARCO'da BM25 iyileştirmeleri                   | Yüksek çevrimdışı üretim; hızlı çevrimiçi         | <sup>[\[50\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-50)[\[51\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-51)</sup> |
| PRF (Rocchio, RLM)        | Geri bildirime dayalı QE | Sorgu (en iyi sonuçlara göre)            | Herhangi biri         | İsteğe bağlı                  | Recall artışı/sapma riskleri                       | \+ ek retrieval geçişi                            | <sup>[\[52\]](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_note-52)</sup>                                                                                                      |

HyDE ve İlgili Yaklaşımların Karşılaştırması

## Ayrıca bakınız

- BM25
- Vektörel gösterimlerle arama
- RAG
- Sözde-ilgili geri bildirim
- BEIR

## Dış bağlantılar

- HyDE deposu: github.com/texttron/hyde.
- Belgelendirme: Haystack — HyDE: docs.haystack.deepset.ai.
- Belgelendirme: LangChain — HyDE Retriever: docs.langchain.com.

## Kaynakça

- 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 IR, 3(4), 333–389. DOI:10.1561/1500000019.

## Notlar

1.  <span id="cite_note-1">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-1) Gao, L.; Ma, X.; Lin, J.; Callan, J. (2023). ‘‘Precise Zero‑Shot Dense Retrieval without Relevance Labels (HyDE)’’. ACL 2023. pp. 1762–1777. DOI:10.18653/v1/2023.acl-long.99. <a href="https://arxiv.org/abs/2212.10496" class="external text" rel="nofollow">arXiv:2212.10496</a></span>
2.  <span id="cite_note-2">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-2) Gao, L. et al. (2023). ACL 2023, §3.2. DOI:10.18653/v1/2023.acl-long.99.</span>
3.  <span id="cite_note-3">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-3) Rocchio, J. (1971). ‘‘Relevance Feedback in Information Retrieval’’. In: Salton, G. (ed.) *The SMART Retrieval System*. Prentice‑Hall, pp. 313–323. ISBN 978‑0138145255.</span>
4.  <span id="cite_note-4">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-4) Lavrenko, V.; Croft, W. B. (2001). ‘‘Relevance‑Based Language Models’’. SIGIR. DOI:10.1145/383952.383972.</span>
5.  <span id="cite_note-5">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-5) Izacard, G. et al. (2021/2022). ‘‘Unsupervised Dense Information Retrieval with Contrastive Learning’’. <a href="https://arxiv.org/abs/2112.09118" class="external text" rel="nofollow">arXiv:2112.09118</a>.</span>
6.  <span id="cite_note-6">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-6) Karpukhin, V. et al. (2020). ‘‘Dense Passage Retrieval for Open‑Domain QA’’. EMNLP. DOI:10.18653/v1/2020.emnlp-main.550.</span>
7.  <span id="cite_note-7">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-7) Thakur, N. et al. (2021). ‘‘BEIR: A Heterogeneous Benchmark for Zero‑shot Evaluation of Information Retrieval Models’’. NeurIPS Datasets Track. <a href="https://arxiv.org/abs/2104.08663" class="external text" rel="nofollow">arXiv:2104.08663</a>.</span>
8.  <span id="cite_note-8">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-8) Gao, L. et al. (2023). DOI:10.18653/v1/2023.acl-long.99.</span>
9.  <span id="cite_note-9">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-9) Milvus Docs. ‘‘Similarity Metrics’’ — при L2‑нормализации векторов внутр. произведение эквивалентно косинусу. URL: <a href="https://milvus.io/docs/v2.2.x/metric.md" class="external free" rel="nofollow">https://milvus.io/docs/v2.2.x/metric.md</a></span>
10. <span id="cite_note-10">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-10) Gao, L.; Ma, X.; Lin, J.; Callan, J. (2023). ‘‘Precise Zero‑Shot Dense Retrieval without Relevance Labels (HyDE)’’. ACL 2023, §3–4. arXiv:2212.10496. DOI:10.18653/v1/2023.acl-long.99.</span>
11. <span id="cite_note-11">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-11) Izacard, G. et al. (2021/2022). ‘‘Unsupervised Dense Information Retrieval with Contrastive Learning (Contriever)’’. arXiv:2112.09118.</span>
12. <span id="cite_note-12">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-12) Gao, L. et al. (2023). Прил. (ablation): влияние числа гипотетических текстов и параметров генерации. arXiv:2212.10496.</span>
13. <span id="cite_note-13">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-13) Gao, L. et al. (2023). DOI:10.18653/v1/2023.acl-long.99.</span>
14. <span id="cite_note-14">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-14) Nogueira, R. et al. (2019). ‘‘Document Expansion by Query Prediction’’ (doc2query). <a href="https://arxiv.org/abs/1904.08375" class="external text" rel="nofollow">arXiv:1904.08375</a>.</span>
15. <span id="cite_note-15">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-15) Nogueira, R.; Lin, J. (2019). ‘‘From doc2query to docTTTTTquery’’ (tech report). <a href="https://cs.uwaterloo.ca/~jimmylin/publications/Nogueira_Lin_2019_docTTTTTquery-v2.pdf" class="external text" rel="nofollow">PDF</a></span>
16. <span id="cite_note-16">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-16) Rocchio, J. (1971); Lavrenko & Croft (2001), см. выше.</span>
17. <span id="cite_note-17">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-17) Nogueira, R.; Cho, K. (2019). ‘‘Passage Re‑ranking with BERT’’. <a href="https://arxiv.org/abs/1901.04085" class="external text" rel="nofollow">arXiv:1901.04085</a>.</span>
18. <span id="cite_note-18">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-18) Khattab, O.; Zaharia, M. (2020). ‘‘ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT’’. SIGIR. DOI:10.1145/3397271.3401075; <a href="https://arxiv.org/abs/2004.12832" class="external text" rel="nofollow">arXiv:2004.12832</a>.</span>
19. <span id="cite_note-19">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-19) Cormack, G. V.; Clarke, C. L. A.; Büttcher, S. (2009). ‘‘Reciprocal Rank Fusion Outperforms Condorcet and Nearly Optimally Combines Rankings’’. SIGIR. DOI:10.1145/1571941.1572114.</span>
20. <span id="cite_note-20">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-20) Gao, L. et al. (2023). Табл. 1. DOI:10.18653/v1/2023.acl-long.99.</span>
21. <span id="cite_note-21">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-21) Izacard, G. et al. (2022); сводные метрики — в Gao et al., 2023, табл. 1. arXiv:2112.09118.</span>
22. <span id="cite_note-22">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-22) Gao, L. et al. (2023). Табл. 1.</span>
23. <span id="cite_note-23">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-23) Karpukhin, V. et al. (2020); сводные — в Gao et al., 2023.</span>
24. <span id="cite_note-24">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-24) Xiong, L. et al. (2021). ICLR. <a href="https://arxiv.org/abs/2007.00808" class="external text" rel="nofollow">arXiv:2007.00808</a>.</span>
25. <span id="cite_note-25">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-25) Thakur, N. et al. (2021); сводные — в Gao et al., 2023, табл. 2. arXiv:2104.08663.</span>
26. <span id="cite_note-26">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-26) Izacard, G. et al. (2022); сводные — в Gao et al., 2023, табл. 2.</span>
27. <span id="cite_note-27">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-27) Gao, L. et al. (2023). Табл. 2.</span>
28. <span id="cite_note-28">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-28) Gao, L. et al. (2023). Табл. 3. DOI:10.18653/v1/2023.acl-long.99.</span>
29. <span id="cite_note-29">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-29) Gao, L. et al. (2023). §4–5.</span>
30. <span id="cite_note-30">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-30) Gao, L. et al. (2023). §4.2–4.3.</span>
31. <span id="cite_note-31">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-31) Gao, L. et al. (2023). §4.1.</span>
32. <span id="cite_note-32">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-32) Haystack Docs. ‘‘Hypothetical Document Embeddings (HyDE)’’ (инженерная справка). <a href="https://docs.haystack.deepset.ai/docs/hypothetical-document-embeddings-hyde" class="external text" rel="nofollow">docs.haystack.deepset.ai</a></span>
33. <span id="cite_note-33">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-33) Gao, L. et al. (2023). Табл. 6.</span>
34. <span id="cite_note-34">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-34) Milvus Docs. ‘‘Similarity Metrics’’.</span>
35. <span id="cite_note-35">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-35) Haystack × Milvus Integration (официальная док.). <a href="https://haystack.deepset.ai/integrations/milvus-document-store" class="external text" rel="nofollow">haystack.deepset.ai</a></span>
36. <span id="cite_note-36">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-36) Nogueira, R.; Cho, K. (2019). arXiv:1901.04085.</span>
37. <span id="cite_note-37">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-37) Khattab, O.; Zaharia, M. (2020). DOI:10.1145/3397271.3401075.</span>
38. <span id="cite_note-38">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-38) Cormack, G. V. et al. (2009). DOI:10.1145/1571941.1572114.</span>
39. <span id="cite_note-39">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-39) Manning, C. D.; Raghavan, P.; Schütze, H. (2008). *Introduction to Information Retrieval*. Cambridge Univ. Press. ISBN 978‑0521865715.</span>
40. <span id="cite_note-40">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-40) Es, S. et al. (2023). ‘‘RAGAS: Automated Evaluation of Retrieval‑Augmented Generation’’. <a href="https://arxiv.org/abs/2309.15217" class="external text" rel="nofollow">arXiv:2309.15217</a>.</span>
41. <span id="cite_note-41">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-41) Gao, L. et al. (2023). §5.</span>
42. <span id="cite_note-42">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-42) Gao, L. et al. (2023). §3.2; §4.1. DOI:10.18653/v1/2023.acl-long.99.</span>
43. <span id="cite_note-43">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-43) Gao, L. et al. (2023). Табл. 3; §4.4.</span>
44. <span id="cite_note-44">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-44) Gao, L. et al. (2023). §4–5.</span>
45. <span id="cite_note-45">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-45) Ouyang, L. et al. (2022). ‘‘Training language models to follow instructions with human feedback (InstructGPT)’’. NeurIPS. <a href="https://arxiv.org/abs/2203.02155" class="external text" rel="nofollow">arXiv:2203.02155</a>.</span>
46. <span id="cite_note-46">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-46) Gao, L. et al. (2023). Табл. 1–2.</span>
47. <span id="cite_note-47">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-47) Robertson, S.; Zaragoza, H. (2009). ‘‘The Probabilistic Relevance Framework: BM25 and Beyond’’. Found. Trends IR. DOI:10.1561/1500000019.</span>
48. <span id="cite_note-48">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-48) Karpukhin, V. et al. (2020). DOI:10.18653/v1/2020.emnlp-main.550.</span>
49. <span id="cite_note-49">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-49) Xiong, L. et al. (2021). <a href="https://arxiv.org/abs/2007.00808" class="external text" rel="nofollow">arXiv:2007.00808</a>.</span>
50. <span id="cite_note-50">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-50) Nogueira, R. et al. (2019). arXiv:1904.08375.</span>
51. <span id="cite_note-51">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-51) Nogueira, R.; Lin, J. (2019). tech report.</span>
52. <span id="cite_note-52">[↑](https://systems-analysis.info/int/Hypothetical_Document_Embeddings_(HyDE)_(TR)#cite_ref-52) Rocchio, J. (1971). SMART; Lavrenko & Croft (2001) SIGIR.</span>