RAG / 检索增强生成 - arXivDaily 专题

2208.05663 2022-08-12 cs.IR 79%

On the Value of Behavioral Representations for Dense Retrieval

Nan Jiang, Dhivya Eswaran, Choon Hui Teo, Yexiang Xue, Yesh Dattatreya, Sujay Sanghavi, Vishy Vishwanathan

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

2012.04584 2022-08-05 cs.CL cs.LG 79%

Distilling Knowledge from Reader to Retriever for Question Answering

Gautier Izacard, Edouard Grave

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

2206.08506 2022-06-20 cs.CL 79%

A Numerical Reasoning Question Answering System with Fine-grained Retriever and the Ensemble of Multiple Generators for FinQA

Bin Wang, Jiangzhou Ju, Yunlin Mao, Xin-Yu Dai, Shujian Huang, Jiajun Chen

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

2205.02303 2022-05-06 cs.IR 79%

Analysing the Robustness of Dual Encoders for Dense Retrieval Against Misspellings

Georgios Sidiropoulos, Evangelos Kanoulas

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments Accepted at the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2022)

2204.13679 2022-04-29 cs.IR cs.LG 79%

Curriculum Learning for Dense Retrieval Distillation

Hansi Zeng, Hamed Zamani, Vishwa Vinay

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments Accepted to SIGIR 2022

2204.10641 2022-04-25 cs.IR 79%

Pre-train a Discriminative Text Encoder for Dense Retrieval via Contrastive Span Prediction

Xinyu Ma, Jiafeng Guo, Ruqing Zhang, Yixing Fan, Xueqi Cheng

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments Accepted to SIGIR 2022

2203.16187 2022-03-31 cs.CL 79%

Auto-MLM: Improved Contrastive Learning for Self-supervised Multi-lingual Knowledge Retrieval

Wenshen Xu, Mieradilijiang Maimaiti, Yuanhang Zheng, Xin Tang, Ji Zhang

专题命中检索器与排序：knowledge retrieval(title,abstract)；分类 cs.CL

Comments 9 pages, 5 figures, 3 tables

2203.08144 2022-03-17 q-fin.ST cs.CL cs.LG cs.SI 79%

DeepTrust: A Reliable Financial Knowledge Retrieval Framework For Explaining Extreme Pricing Anomalies

Pok Wah Chan

专题命中检索器与排序：knowledge retrieval(title,abstract)；分类 cs.CL

Comments 72 pages

2111.13957 2021-11-30 cs.IR 79%

Interpreting Dense Retrieval as Mixture of Topics

Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang, Shaoping Ma

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

2109.04014 2021-09-10 cs.CL 79%

Weakly-Supervised Visual-Retriever-Reader for Knowledge-based Question Answering

Man Luo, Yankai Zeng, Pratyay Banerjee, Chitta Baral

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

Comments accepted at EMNLP 2021

2108.03937 2021-08-10 cs.IR 79%

DoSSIER@COLIEE 2021: Leveraging dense retrieval and summarization-based re-ranking for case law retrieval

Sophia Althammer, Arian Askari, Suzan Verberne, Allan Hanbury

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments Published in COLIEE 2021

2107.07773 2021-07-19 cs.IR 79%

More Robust Dense Retrieval with Contrastive Dual Learning

Yizhi Li, Zhenghao Liu, Chenyan Xiong, Zhiyuan Liu

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments Accepted by ICTIR 2021

2106.09983 2021-06-21 cs.CL 79%

Weakly Supervised Pre-Training for Multi-Hop Retriever

Yeon Seonwoo, Sang-Woo Lee, Ji-Hoon Kim, Jung-Woo Ha, Alice Oh

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

Comments ACL-Findings 2021

2104.08723 2021-04-20 cs.CL 79%

News Meets Microblog: Hashtag Annotation via Retriever-Generator

Xiuwen Zheng, Dheeraj Mekala, Amarnath Gupta, Jingbo Shang

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

2104.08051 2021-04-19 cs.IR 79%

Optimizing Dense Retrieval Model Training with Hard Negatives

Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang, Shaoping Ma

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

Comments To be published in SIGIR2021

2104.05883 2021-04-14 cs.CL 79%

Multi-Step Reasoning Over Unstructured Text with Beam Dense Retrieval

Chen Zhao, Chenyan Xiong, Jordan Boyd-Graber, Hal Daumé

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.CL

Comments NAACL 2021

2009.12756 2021-02-23 cs.CL 79%

Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval

Wenhan Xiong, Xiang Lorraine Li, Srini Iyer, Jingfei Du, Patrick Lewis, William Yang Wang, Yashar Mehdad, Wen-tau Yih, Sebastian Riedel, Douwe Kiela, Barlas Oğuz

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.CL

2010.06189 2020-10-28 cs.CL 79%

X-FACTR: Multilingual Factual Knowledge Retrieval from Pretrained Language Models

Zhengbao Jiang, Antonios Anastasopoulos, Jun Araki, Haibo Ding, Graham Neubig

专题命中检索器与排序：knowledge retrieval(title,abstract)；分类 cs.CL

Comments EMNLP 2020

2010.10999 2020-10-22 cs.CL 79%

Is Retriever Merely an Approximator of Reader?

Sohee Yang, Minjoon Seo

专题命中检索器与排序：retriever(title,abstract)；分类 cs.CL

2010.10469 2020-10-21 cs.IR 79%

Learning To Retrieve: How to Train a Dense Retrieval Model Effectively and Efficiently

Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Min Zhang, Shaoping Ma

专题命中检索器与排序：dense retrieval(title,abstract)；分类 cs.IR

2608.16394 2026-08-18 cs.AI cs.IR 新提交 79%

Think Inside the Chunk: RegulaRAG for Regulation-Compliant Scenario Generation using LLMs: A Case Study of UN Regulation No. 152

在块内思考：使用大语言模型生成符合法规的场景的RegulaRAG——以联合国第152号法规为例

Vahid Zolfaghari, Nenad Petrovic, AndrÉ Schamschurko, Alois Knoll

机构 * Technical University of Munich（慕尼黑工业大学）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.IR、cs.AI

AI总结针对LLMs难以结合冗长分层标准的问题，提出RegulaRAG流水线，经实验其在UN R152数据集上元分数最高且鲁棒性强，优于基线系统。

2607.22067 2026-08-18 cs.CL cs.AI 版本更新 79%

Multimodal Language Models Benchmarked Against the NRC Reactor Operator Licensing Examination: Fine-Tuning and Retrieval Strategies

基于美国核管理委员会反应堆操作员执照考试的多模态语言模型微调与检索策略基准测试

Isak Hwang, Yoon Pyo Lee, Syed Bahauddin Alam

机构 * organization= Department of Nuclear Engineering, Hanyang University , addressline= 222 Wangsimni-ro , postcode= 04763 , state= Seongdong-gu , city= Seoul , country= South Korea ； organization= The Grainger College of Engineering, Nuclear, Plasma \& Radiological Engineering, University of Illinois Urbana-Champaign , city= Urbana , state= IL , country= USA

专题命中检索器与排序：RAG(summary_cn,abstract_cn)；分类 cs.CL、cs.AI

AI总结该研究针对美国核管理委员会反应堆操作员执照考试，评估310亿参数多模态模型应用核知识的能力，通过对比基础模型与多种微调及检索配置，发现固定大小分块RAG的SFT配置表现最佳，并揭示了分块策略规律及RAFT与SFT的性能差异。

2604.22207 2026-08-05 cs.SE cs.AI cs.CL 版本更新 79%

Evaluating LLM-Based Goal Extraction in Requirements Engineering: Prompting Strategies and Their Limitations

评估基于大语言模型的目标提取在需求工程中的应用：提示策略及其局限性

Anna Arnaudo, Riccardo Coppola, Maurizio Morisio, Flavio Giobergia, Andrea Bioddo, Angelo Bongiorno, Luca Dadone

机构 * Department of Control and Computer Engineering（控制与计算机工程系）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.CL、cs.AI

AI总结本文探讨了通过三个阶段自动提取功能目标以实现目标导向的需求工程，提出基于工程提示的LLM链，实验表明反馈循环机制在零样本学习中表现更优，但提示策略仍是性能限制因素。

Comments 11 pages, 1 figure. This contribution will be published in the conference proceedings of EASE 2026 Conference (https://conf.researchr.org/home/ease-2026/prompt-se-2026)

2607.04281 2026-07-07 cs.CL cs.AI 新提交 79%

Risk-Constrained Freshness-Aware Semantic Caching for Open-Web Retrieval-Augmented LLMs

面向开放网络检索增强语言模型的风险约束新鲜度感知语义缓存

Muhammad Mansoor, Tahir Ahmad, Yeo-Chan Yoon

机构 * Jeju National University（济州国立大学）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.CL、cs.AI

AI总结研究针对开放网络证据时变新鲜度，提出三层语义缓存FreshCache，将缓存重用视为风险约束时间推理问题，给出评估方法并引入基准测试，实验表明其在节省搜索API及降低陈旧错误率方面效果良好。

2607.00023 2026-07-02 cs.IR cs.AI 新提交 79%

Aligning Sentence Embeddings to Human Concepts via Sparse Autoencoders

通过稀疏自编码器将句子嵌入与人类概念对齐

Wonseok Shin, Songkuk Kim

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.IR、cs.AI

AI总结提出使用Top-k稀疏自编码器解耦句子嵌入，使其特征与语义、句法等人类概念对齐，并通过激活引导机制实现可解释的检索重排序。

2606.29824 2026-06-30 cs.CL cs.AI 79%

Neural Procedural Memory: Empowering LLM Agents with Implicit Activation Steering

神经程序记忆：通过隐式激活引导赋予LLM智能体能力

Chengfeng Zhao, Yuqiao Tan, Shizhu He, Yequan Wang, Jun Zhao, Kang Liu

机构 * Institute of Automation, CAS（中国科学院自动化研究所）； University of Chinese Academy of Sciences（中国科学院大学）； Beijing Academy of Artificial Intelligence（北京人工智能研究院）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.CL、cs.AI

AI总结提出神经程序记忆（NPM），一种无训练框架，通过隐式激活引导而非显式指令表示智能体记忆，从历史对比经验中提取程序技能为激活空间中的引导向量，直接激活任务相关神经机制以指导执行。

2606.26157 2026-06-26 cs.IR cs.AI 新提交 79%

Reducing Redundancy in Whole-Slide Image Patching for Scalable Indexing and Retrieval

减少全切片图像分块冗余以实现可扩展索引与检索

Jialiang Geng, Ghazal Alabtah, Saghir Alfasly, Wataru Uegami, H. R. Tizhoosh

机构 * KIMIA Lab Dept. of Artificial Intelligence & Informatics（KIMIA实验室人工智能与信息学系）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.IR、cs.AI

AI总结提出ARReST框架，通过识别并剪除跨类别判别贡献小的对立块，在保持检索精度的同时显著压缩WSI索引存储（3%-60%），实现可扩展、低成本的病理图像检索。

2606.15591 2026-06-16 cs.AI cs.CL cs.MA 新提交 79%

Agentic Retrieval and Reinforcement Learned Equation Chains: A Controlled Generation Framework for Complex and Novel Physics Word Problems

智能检索与强化学习方程链：面向复杂新颖物理文字题的可控生成框架

Tirthankar Mittra

机构 * University of California, Berkeley（加州大学伯克利分校）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.CL、cs.AI

AI总结提出ARVRE两阶段框架，通过离线时序差分学习构建有效物理方程链，结合智能检索增强生成控制问题结构与难度，再由大语言模型生成自然语言问题，实现复杂、新颖且可解的物理文字题生成。

2510.18355 2026-06-16 cs.CL cs.HC cs.IR 79%

KrishokBondhu: A Retrieval-Augmented Voice-Based Agricultural Advisory Call Center for Bengali Farmers

KrishokBondhu：一种基于检索增强的语音农业咨询呼叫中心，面向孟加拉语农民

Mohd Ruhul Ameen, Akif Islam, Farjana Aktar, M. Saifuzzaman Rafat

机构 * University of California, Berkeley（加州大学伯克利分校）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.IR、cs.CL

AI总结本文提出KrishokBondhu，一种基于检索增强生成框架的语音农业咨询平台，通过OCR处理农业手册等资料，结合大语言模型生成回答，实现孟加拉语农民的实时农业指导。

Comments Accepted at the 2026 IEEE 2nd International Conference on Quantum Photonics, Artificial Intelligence and Networking (QPAIN 2026)

Journal ref 2026 IEEE 2nd International Conference on Quantum Photonics, Artificial Intelligence & Networking (QPAIN)

2606.11198 2026-06-11 cs.CL cs.AI 新提交 79%

The Structural Attention Tax: How Retrieval Format Hijacks In-Context Learning Independent of Content

结构注意力税：检索格式如何劫持上下文学习而与内容无关

Yuqi Zhang, Di Zhang

机构 * Xi’an Jiaotong-Liverpool University（西交利物浦大学）

专题命中检索器与排序：RAG(abstract,abstract_cn)；retrieval-augmented generation(abstract)；分类 cs.CL、cs.AI

AI总结研究发现知识图谱三元组因其格式结构比自然语言吸引2-3倍注意力，压缩演示注意力达42%，并提出了分解注意力为语义与结构成分的框架及缓解策略。

Comments 10 pages, 5 figures

1. 检索器与排序 4551 篇

On the Value of Behavioral Representations for Dense Retrieval

Distilling Knowledge from Reader to Retriever for Question Answering

A Numerical Reasoning Question Answering System with Fine-grained Retriever and the Ensemble of Multiple Generators for FinQA

Analysing the Robustness of Dual Encoders for Dense Retrieval Against Misspellings

Curriculum Learning for Dense Retrieval Distillation

Pre-train a Discriminative Text Encoder for Dense Retrieval via Contrastive Span Prediction

Auto-MLM: Improved Contrastive Learning for Self-supervised Multi-lingual Knowledge Retrieval

DeepTrust: A Reliable Financial Knowledge Retrieval Framework For Explaining Extreme Pricing Anomalies

Interpreting Dense Retrieval as Mixture of Topics

Weakly-Supervised Visual-Retriever-Reader for Knowledge-based Question Answering

DoSSIER@COLIEE 2021: Leveraging dense retrieval and summarization-based re-ranking for case law retrieval

More Robust Dense Retrieval with Contrastive Dual Learning

Weakly Supervised Pre-Training for Multi-Hop Retriever

News Meets Microblog: Hashtag Annotation via Retriever-Generator

Optimizing Dense Retrieval Model Training with Hard Negatives

Multi-Step Reasoning Over Unstructured Text with Beam Dense Retrieval

Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval

X-FACTR: Multilingual Factual Knowledge Retrieval from Pretrained Language Models

Is Retriever Merely an Approximator of Reader?

Learning To Retrieve: How to Train a Dense Retrieval Model Effectively and Efficiently

Think Inside the Chunk: RegulaRAG for Regulation-Compliant Scenario Generation using LLMs: A Case Study of UN Regulation No. 152

Multimodal Language Models Benchmarked Against the NRC Reactor Operator Licensing Examination: Fine-Tuning and Retrieval Strategies

Evaluating LLM-Based Goal Extraction in Requirements Engineering: Prompting Strategies and Their Limitations

Risk-Constrained Freshness-Aware Semantic Caching for Open-Web Retrieval-Augmented LLMs

Aligning Sentence Embeddings to Human Concepts via Sparse Autoencoders

Neural Procedural Memory: Empowering LLM Agents with Implicit Activation Steering

Reducing Redundancy in Whole-Slide Image Patching for Scalable Indexing and Retrieval

Agentic Retrieval and Reinforcement Learned Equation Chains: A Controlled Generation Framework for Complex and Novel Physics Word Problems

KrishokBondhu: A Retrieval-Augmented Voice-Based Agricultural Advisory Call Center for Bengali Farmers

The Structural Attention Tax: How Retrieval Format Hijacks In-Context Learning Independent of Content