发表机构
University of Primorska; InnoRenew CoE; Research Centre of the Slovenian Academy of Sciences and Arts; The Fran Ramovš Institute（普里莫尔斯卡大学; InnoRenew CoE; 斯洛文尼亚科学与艺术研究院研究中心; 弗兰·拉莫夫什研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出假设提示嵌入（HyPE）框架，将假设内容生成移至索引阶段，提升RAG的检索对齐度与效率，在六个数据集上实现检索精度和召回率的显著提升且兼容多种RAG技术
AI 中文摘要
检索增强生成（RAG）系统将检索机制与生成式语言模型相结合，以提升响应的准确性和相关性。然而，在检索增强系统中，弥合用户查询与文档文本中相关信息之间的风格差距仍是一个持续存在的挑战，通常通过运行时解决方案（如假设文档嵌入（HyDE））来解决，这些方案试图提升对齐度，但会在查询时引入额外的计算开销。为应对这些挑战，我们提出假设提示嵌入（HyPE）框架，该框架将假设内容的生成从查询阶段转移到索引阶段。通过为每个数据块预先计算多个假设提示，并将该提示嵌入到数据块中，HyPE将检索转化为问答匹配任务，绕过了运行时合成答案生成的需求。该方法不会引入延迟，同时还增强了查询与相关上下文之间的对齐度。我们在六个常用数据集上的实验结果表明，与标准方法相比，HyPE可将检索上下文精度提升多达42个百分点，主张召回率提升多达45个百分点，且与重排序、多向量检索、查询分解及其他RAG进展兼容
英文摘要
Retrieval-Augmented Generation (RAG) systems synergize retrieval mechanisms with generative language models to enhance the accuracy and relevance of responses. However, bridging the style gap between user queries and relevant information in document text remains a persistent challenge in retrieval-augmented systems, often addressed by runtime solutions (e.g., Hypothetical Document Embeddings (HyDE)) that attempt to improve alignment but introduce extra computational overhead at query time. To address these challenges, we propose Hypothetical Prompt Embeddings (HyPE), a framework that shifts the generation of hypothetical content from query time to the indexing phase. By precomputing multiple hypothetical prompts for each data chunk and embedding the chunk in place of the prompt, HyPE transforms retrieval into a question-question matching task, bypassing the need for runtime synthetic answer generation. This approach does not introduce latency but also strengthens the alignment between queries and relevant context. Our experimental results on six common datasets show that HyPE can improve retrieval context precision by up to 42 percentage points and claim recall by up to 45 percentage points, compared to standard approaches, while remaining compatible with re-ranking, multi-vector retrieval, query decomposition, and other RAG advancements
Comments10 pages, 8 figures, 5 tables. Published in IEEE Access
Journal refIEEE Access, vol. 13, pp. 129952-129961, 2025
DOI：10.1109/ACCESS.2025.3589499
