Dense Retrieval

الاسترجاع الكثيف

أسلوب استرجاع معلومات يُمثِّل النصوص والاستعلامات كمتجهات كثيفة (ذات قيم حقيقية) في فضاء مشترك، ثم يقيس التشابه الدلالي بينها عبر الضرب النقطي أو تشابه جيب التمام، بدلاً من مطابقة الكلمات المفتاحية.

An information retrieval approach that represents queries and documents as dense (real-valued) vectors in a shared space and measures semantic similarity via dot product or cosine similarity, rather than keyword matching.

Also translated asالاسترجاع الدلالي الكثيف، الاسترجاع بالتمثيلات المتجهية

First appears in this corpus in: Okapi at TREC-3 (1994)

Appears in these papers

Atlas: Few-shot Learning with Retrieval Augmented Language Models2023 in the sky ✦
Atlas: Few-shot Learning with Retrieval Augmented Language Models2023 in the sky ✦
Okapi at TREC-31994 in the sky ✦
Okapi at TREC-31994 in the sky ✦
ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT2020 in the sky ✦
Contriever: Unsupervised Dense Information Retrieval with Contrastive Learning2022 in the sky ✦
Contriever: Unsupervised Dense Information Retrieval with Contrastive Learning2022 in the sky ✦
Dense Passage Retrieval for Open-Domain Question Answering2020 in the sky ✦
Dense Passage Retrieval for Open-Domain Question Answering2020 in the sky ✦
Billion-Scale Similarity Search with GPUs2017 in the sky ✦
HyDE: Precise Zero-Shot Dense Retrieval without Relevance Labels2022 in the sky ✦
HyDE: Precise Zero-Shot Dense Retrieval without Relevance Labels2022 in the sky ✦
REALM: Retrieval-Augmented Language Model Pre-Training2020 in the sky ✦
RETRO: Improving Language Models by Retrieving from Trillions of Tokens2022 in the sky ✦
Sentence-BERT: Sentence Embeddings Using Siamese BERT-Networks2019 in the sky ✦