RAG / 检索增强生成

Mapping Similarity Spaces across Embedding Models with Synthetic Query Probing

通过合成查询探测映射嵌入模型间的相似度空间

Marcin Rozmus, Peter van der Putten

机构 * Pegasystems（pegasystems公司） ； Leiden University（莱顿大学） 纠错

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract_cn)；分类 cs.CL

AI总结 本研究针对嵌入模型间相似度分数不可直接比较的问题，提出合成查询探测方法，通过学习分数分布映射实现跨模型相似度空间对齐，提升阈值可移植性，为嵌入可比性分析提供可扩展框架。

Comments Accepted for 29th International Conference on Discovery Science, October 5-9, 2026, Mainz, Germany

RCEM: Robust Conversational Search EMbedder in Distributional Shift

RCEM：配备查询重写技能的嵌入器，用于分布偏移下的鲁棒对话搜索

Kilho Son, Paul Hsu, Cha Zhang, Dinei Florencio

机构 * Microsoft（微软） 纠错

专题命中 向量检索 ：RAG(abstract_cn)；dense retrieval(abstract)；分类 cs.CL

AI总结 提出RCEM模型，通过将LLM的查询重写能力蒸馏到嵌入模型中，实现无需显式重写的上下文感知检索，在分布偏移下提升鲁棒性。

When Does Embedding Magnitude Matter? A Cross-Task Functional-Symmetry Framework

嵌入量何时重要？一种跨任务功能对称框架

Xincan Feng, Taro Watanabe

机构 * Nara Institute of Science and Technology（奈良科学技术研究所） 纠错

专题命中 向量检索 ：RAG(abstract,abstract_cn)；分类 cs.IR

AI总结 本文提出一种2x2框架，独立控制查询和文档侧归一化，发现QNorm和DNorm在跨任务中优于余弦和点积，揭示文档量影响推理分数，查询量调节训练梯度，Fisher信息矩阵条件数预测归一化侧。

Comments Preliminary work. Under review

FGIM: a Fast Graph-based Indexes Merging Framework for Approximate Nearest Neighbor Search

FGIM: 一种基于图的索引合并框架用于近邻搜索

Zekai Wu, Jiabao Jin, Peng Cheng, Xiaoyao Zhong, Lei Chen, Yongxin Tong, Zhitao Shen, Jingkuan Song, Heng Tao Shen, Xuemin Lin

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.DB

AI总结 本文提出FGIM框架，通过三种核心技术提升图索引合并效率，实验显示其在多种近邻搜索方法中实现显著加速且保持性能。

Comments 27 pages, accepted by SIGMOD 2026

Embedding-Based Context-Aware Reranker

基于嵌入的上下文感知重排序器

Ye Yuan, Mohammad Amin Shabani, Siqi Liu

机构 * McGill University（麦吉尔大学） ； Mila - Quebec AI Institute（魁北克人工智能研究所） ； RBC Borealis 纠错

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL

AI总结 本文提出EBCAR，一种基于嵌入的上下文感知重排序器，通过增强跨段落理解提升信息检索的准确性和效率。

Comments Accepted by ICLR 2026

Welfarist Formulations for Diverse Similarity Search

效用主义方法用于多样相似性搜索

Siddharth Barman, Nirjhar Das, Shivam Gupta, Kirankumar Shiragur

机构 * Indian Institute of Science（印度科学研究所） ； Microsoft Research India（微软印度研究院） 纠错

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.IR

AI总结 本文提出基于效用主义的NNS方法，通过适应性平衡相关性与多样性，提升搜索结果的多样性同时保持高相关性。

SpANNS: Optimizing Approximate Nearest Neighbor Search for Sparse Vectors Using Near Memory Processing

SpANNS: 通过近内存处理优化稀疏向量的近似最近邻搜索

Tianqi Zhang, Flavio Ponzina, Tajana Rosing

专题命中 向量检索 ：vector search(abstract)；hybrid retrieval(abstract)；分类 cs.DB

AI总结 SpANNS通过近内存处理架构优化稀疏向量的近似最近邻搜索，显著提升搜索效率。

Adaptation of Embedding Models to Financial Filings via LLM Distillation

通过LLM蒸馏适应金融文件的嵌入模型

Eliot Brenner, Dominic Seyler, Manjunath Hegde, Andrei Simion, Koustuv Dasgupta, Bing Xiang

专题命中 向量检索 ：RAG(abstract)；retriever(abstract)；分类 cs.CL

AI总结 本文通过LLM蒸馏方法，提升金融领域检索性能，实现MRR@5和DCG@5的显著提升。

Comments In proceedings of LLM-Finance 2025 : The 2nd IEEE International Workshop on Large Language Models for Finance

Approximate Diverse $k$-nearest Neighbor Search in Vector Database

Jiachen Zhao, Xiao Yan, Eric Lo

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.DB

Query Drift Compensation: Enabling Compatibility in Continual Learning of Retrieval Embedding Models

Dipam Goswami, Liying Wang, Bartłomiej Twardowski, Joost van de Weijer

机构 * Computer Vision Center, Barcelona, Spain（巴塞罗那计算机视觉中心） ； Department of Computer Science, Universitat Autònoma de Barcelona, Spain（巴塞罗那自治大学计算机科学系） ； IDEAS Research Center, Warsaw, Poland（华沙IDEAS研究中心） 纠错

专题命中 向量检索 ：retrieval augmented generation(abstract)；dense retrieval(abstract)；分类 cs.IR

Comments Accepted at CoLLAs 2025

WebANNS: Fast and Efficient Approximate Nearest Neighbor Search in Web Browsers

Mugeng Liu, Siqi Zhong, Qi Yang, Yudong Han, Xuanzhe Liu, Yun Ma

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.IR

Comments SIGIR 2025

Beyond instruction-conditioning, MoTE: Mixture of Task Experts for Multi-task Embedding Models

Miguel Romero, Shuoyang Ding, Corey D. Barret, Georgiana Dinu, George Karypis

机构 * Amazon（亚马逊公司） ； University of Minnesota（明尼苏达大学） 纠错

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL

GEM: Empowering LLM for both Embedding Generation and Language Understanding

Caojin Zhang, Qiang Zhang, Ke Li, Sai Vidyaranya Nuthalapati, Benyu Zhang, Jason Liu, Serena Li, Lizhu Zhang, Xiangjun Fan

机构 * Meta Inc（Meta公司） 纠错

专题命中 向量检索 ：retrieval augmented generation(abstract)；RAG(abstract)；分类 cs.CL

Towards A Generalist Code Embedding Model Based On Massive Data Synthesis

Chaofan Li, Jianlyu Chen, Yingxia Shao, Defu Lian, Zheng Liu

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.IR

Decentralizing AI Memory: SHIMI, a Semantic Hierarchical Memory Index for Scalable Agent Reasoning

Tooraj Helmi

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.AI

Operational Advice for Dense and Sparse Retrievers: HNSW, Flat, or Inverted Indexes?

Jimmy Lin

专题命中 向量检索 ：dense retrieval(abstract)；vector search(abstract)；分类 cs.IR

Multi-Lingual Malaysian Embedding: Leveraging Large Language Models for Semantic Representations

Husein Zolkepli, Aisyah Razak, Kamarul Adha, Ariff Nazhan

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL

Novel Preprocessing Technique for Data Embedding in Engineering Code Generation Using Large Language Model

Yu-Chen Lin, Akhilesh Kumar, Norman Chang, Wenliang Zhang, Muhammad Zakir, Rucha Apte, Haiyang He, Chao Wang, Jyh-Shing Roger Jang

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)；分类 cs.CL

DIVE: Embedding Compression via Self-Limiting Gradient Updates

DIVE: 通过自限制梯度更新实现嵌入压缩

Dongfang Zhao

机构 * University of Washington Tacoma School of Engineering and Technology（华盛顿大学塔可姆分校工程与技术学院） 纠错

专题命中 向量检索 ：vector search(abstract)；分类 cs.IR、cs.CL、cs.AI

AI总结 本文提出DIVE方法，通过自限制的三元组损失和头级NT-Xent对比损失解决嵌入压缩中因标注数据稀缺导致的过拟合问题，提升了检索性能。

ANN Search: Recall What Matters

ANN搜索：召回真正重要的

Dimitris Dimitropoulos, Nikos Mamoulis

机构 * University of Ioannina（伊奥尼亚大学） ； Archimedes, Athena RC（阿基米德，雅典RC） 纠错

专题命中 向量检索 ：retrieval-augmented generation(abstract)；分类 cs.IR、cs.AI、cs.DB

AI总结 本文提出用逆近似比1/Ratio@k替代Recall@k来评估近似最近邻搜索质量，实验表明前者能更准确反映实际效用并降低计算开销。

Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini

Gemini Embedding 2：来自Gemini的原生多模态嵌入模型

Madhuri Shanbhogue, Zhe Li, Shanfeng Zhang, Gustavo Hernández Ábrego, Shih-Cheng Huang, Aashi Jain, Daniel Salz, Sonam Goenka, Chaitra Hegde, Ji Ma, Feiyang Chen, Jiaxing Wu, Tanmaya Dabral, Babak Samari, Kevin Poulet, Daniel Cer, Kaifeng Chen, Paul Suganathan, Hui Hui, Jovan Andonov, Philippe Schlattner, Jay Han, Iftekhar Naim, Wing Lowe, Vladimir Pchelin, Albert Yang, Yi-Ting Chen, Zhongli Ding, Grace Zhang, Georg Heigold, Yichang Chen, Antoine Reveillon, Brendan Mccloskey, Wenlei Zhou, Dahun Kim, Rui Meng, Emma Wang, Jack Zheng, Halley Fede, Zhen Yang, Keegan Mosley, Brian Potetz, Sahil Dua, Henrique Schechter Vera, Shen Gao, Hesen Zhang, Andreas Hess, Hengxuan Ying, Alberto Montes, Karan Gill, Min Choi, Sebastian Russo, Anja Hauth, Jinhyuk Lee, Michael Boratko, Megan Barnes, Vikram Rao, Claudiu Musat, Cyril Allauzen, Ehsan Variani, Shankar Kumar, Tom Bagby, Junyi Jiao, Yang Gu, Tengxin Li, Ayush Agrawal, Roberto Santana, Dev Nath, Stephen Karukas, Shuoxuan Han, Lucia Loher, Alice Twu, Nidhi Vyas, Siddharth Bhai, Frank Palma Gomez, Wangyuan Zhang, Chaoren Liu, Jizheng Yang, Steve Qiu, Shijie Zhang, Sujay Kulkarni, Sascha Rothe, Sean Nakamoto, Raphael Hoffmann, Zach Gleicher, Yunhsuan Sung, Qin Yin, Tom Duerig, Mojtaba Seyedhosseini

机构 * Gemini Report（Gemini 报告） 纠错

专题命中 向量检索 ：RAG(abstract,abstract_cn)

AI总结 提出原生多模态嵌入模型Gemini Embedding 2，通过多任务多阶段对比学习统一视频、音频、图像和文本的表示空间，在单模态、跨模态和多模态检索任务上达到最先进性能。

FusedANN: Convexified Hybrid ANN via Attribute-Vector Fusion

Alireza Heidari, Wei Zhang, Ying Xiong

机构 * Huawei Technologies Ltd（华为技术有限公司） 纠错

专题命中 向量检索 ：vector search(abstract)；分类 cs.IR、cs.AI、cs.DB

Comments 62 pages,12 figures

DRIM-ANN: An Approximate Nearest Neighbor Search Engine based on Commercial DRAM-PIMs

Mingkai Chen, Tianhua Han, Cheng Liu, Shengwen Liang, Kuai Yu, Lei Dai, Ziming Yuan, Ying Wang, Lei Zhang, Huawei Li, Xiaowei Li

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)

Cosmos: A CXL-Based Full In-Memory System for Approximate Nearest Neighbor Search

Seoyoung Ko, Hyunjeong Shim, Wanju Doh, Sungmin Yun, Jinin So, Yongsuk Kwon, Sang-Soo Park, Si-Dong Roh, Minyong Yoon, Taeksang Song, Jung Ho Ahn

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)

Comments 4 pages, 5 figures, to appear at IEEE Computer Architecture Letters

LoRANN: Low-Rank Matrix Factorization for Approximate Nearest Neighbor Search

Elias Jääsaari, Ville Hyvönen, Teemu Roos

专题命中 向量检索 ：retrieval-augmented generation(abstract)；RAG(abstract)

Comments Accepted to NeurIPS 2024

NDSEARCH: Accelerating Graph-Traversal-Based Approximate Nearest Neighbor Search through Near Data Processing

Yitu Wang, Shiyu Li, Qilin Zheng, Linghao Song, Zongwang Li, Andrew Chang, Hai "Helen" Li, Yiran Chen

专题命中 向量检索 ：retrieval augmented generation(abstract)；RAG(abstract)

Jina Embeddings: A Novel Set of High-Performance Sentence Embedding Models

Michael Günther, Louis Milliken, Jonathan Geuter, Georgios Mastrapas, Bo Wang, Han Xiao

专题命中 向量检索 ：dense retrieval(abstract)；分类 cs.IR、cs.CL、cs.AI

Comments 9 pages, 2 page appendix

Vec2Vec: A Compact Neural Network Approach for Transforming Text Embeddings with High Fidelity

Andrew Kean Gao

专题命中 向量检索 ：vector search(abstract)；分类 cs.IR、cs.CL、cs.AI

Comments 14 pages, 6 figures, 5 tables

Quick Dense Retrievers Consume KALE: Post Training Kullback Leibler Alignment of Embeddings for Asymmetrical dual encoders

Daniel Campos, Alessandro Magnani, ChengXiang Zhai

专题命中 向量检索 ：dense retrieval(abstract)；分类 cs.IR、cs.CL、cs.AI

Comments SustaiNLP2023 @ ACL 2023, 8 pages, 4 figures, 30 tables

Efficient Privacy-Preserving Range Filtered Approximate Nearest Neighbor Search

高效的隐私保护范围过滤近似最近邻搜索

Haoyu Wang, Yandi Zhang, Jiadong Xie, Yingfan Liu, Hui Li, Jeffrey Xu Yu, Jiangtao Cui

专题命中 向量检索 ：vector search(abstract)；分类 cs.IR、cs.DB

AI总结 本研究针对外包加密向量数据库的隐私保护需求，提出了分离范围定位与加密向量搜索的过滤-精调RFANNS方案，实验验证其在QPS-召回率权衡上优于现有安全适配方案。

Comments According to the best of our knowledge, this work is the first attempt to study privacy-preserving range-filterd ANN search problem. This is the early version of the work that is still in progress