#retrieval-augmented generation
42 resultsUnderstanding Is Done Early: A Depth Division of Labor in Large Language Models and Its Use for Unbounded-Context Memory
Hanzuo Liu, Xuan Qi, Chunyu Liu +6
The paper proposes CoMem, a method that stores intermediate transformer layer states as memory to enable efficient long‑context retrieval, showing that using lower‑mid layers for c…
GLM-RAG: Graph Language Models for Graph-Based Retrieval-Augmented Generation
Maya Arseven, Anette Frank, Beni Egressy +2
The paper introduces a graph language model (GLM) based retriever for retrieval-augmented generation over knowledge graphs and compares it with GNN‑based and vector‑search retrieve…
DualG-MRAG: Decoupling Macro-Reasoning and Micro-Matching for Multimodal Retrieval-Augmented Generation
Jiacheng Tao, Qingyun Sun, Haonan Yuan +2
The paper introduces DualG-MRAG, a framework that separates global reasoning and fine-grained evidence matching using macro and micro graphs to improve multimodal retrieval-augment…
CMT-RAG: Complementary Memory Traces for Multi-turn Multi-hop RAG
Lang Zhou, Yingjian Chen, Shuxuan Li +2
The paper proposes CMT-RAG, a framework that stores structured sub-question reasoning traces as memory to improve multi-turn, multi-hop retrieval‑augmented generation, and introduc…
When Knowledge Changes: Metamorphic Testing of RAG Systems with Mutations
Jinhan Kim, Samuele Pasini, Paolo Tonella
The paper proposes a metamorphic testing framework that assesses how Retrieval‑Augmented Generation (RAG) systems behave when their underlying document corpora change, using a taxo…
RAG-HAR+: Towards Cost-Efficient LLM-Based Human Activity Recognition for Edge Deployment
Hansi Karunarathna, Nirhoshan Sivaroopan, Chamara Madarasingha +2
The paper presents RAG-HAR+, a cost‑efficient system that combines retrieval of similar sensor data with large language model (LLM) assistance to recognize human activities on edge…
VITAL-RAG: Invariance Race for Context Allocation in Coding Agents
Zijian Lu, Yonghua Lu, Mingcai Chen +4
The paper introduces VITAL-RAG, a method for coding agents that groups retrieved code fragments by their original code object and selectively includes only those that add new task-…
Hierarchical Reranking for Scalable Financial RAG System
Joohyun Lee, Sungwoo Hong
The paper introduces a Hierarchical Reranker framework for Retrieval-Augmented Generation that improves retrieval accuracy and handles long contexts when processing large-scale fin…
BM25 Wins at Scale: A Scaling Study of Retrieval-Augmented Generation Paradigms
Pengyu Wang, Benfeng Xu, Shaohan Wang +5
The paper conducts a controlled scaling study of various retrieval-augmented generation methods and finds that BM25 becomes the most accurate and cost‑effective approach once the c…
From Backlog Items to Security Guidance: Towards Continuous Security Compliance
Ignacio GarcÃa Núñez, Florian Angermeir, Fabiola Moyón Constante
The paper introduces an NLP-based system that automatically identifies security‑relevant items in software development backlogs and links them to appropriate security requirements,…
Models for minimalist RAG: B1ade 335M Embedding and 1B Parameter Small Language Models
Shreyas Subramanian, Mecit Gungor, Vikram Elango
The paper presents B1ade, a resource‑efficient retrieval‑augmented generation system that combines a 335M parameter embedding model built by fusing five pretrained encoders with a…
LayerRAG-Bench: A Cross-Layer Reliability Benchmark for Agentic Retrieval-Augmented Generation
Musa Shams
The paper presents LayerRAG-Bench, a benchmark that evaluates the reliability of agentic retrieval-augmented generation systems across multiple layers such as evidence, tool contra…
Beyond Self-Knowledge: Propagating Uncertainty Across Reasoning and Retrieval in LLMs
Chandan Kumar Sah, Xiaoli Lian, Li Zhang
The paper introduces BeyondUncertainty, a method that uses confidence estimates from black‑box language models to decide whether to retrieve external evidence for question answerin…
RAGuard: A Layered Defense Framework for Retrieval-Augmented Generation Systems Against Data Poisoning
Pushkal Kumar, Tucker Nielson, Tanish Kolhe +2
The paper introduces RAGuard, a two‑layer defense for retrieval‑augmented generation systems that combats factual corpus‑poisoning by adversarially fine‑tuning the retriever and ap…
When Machine Unlearning Meets Retrieval-Augmented Generation (RAG): Keep Secret or Forget Knowledge?
Shang Wang, Tianqing Zhu, Dayong Ye +1
The paper proposes a lightweight method to make large language models forget specific information by altering the external knowledge base of Retrieval‑Augmented Generation systems,…
Digital Pantheon: Simulating and Auditing Coalition Formation with LLM Agents
Dylan Van Mulders, Matthias Bogaert, Dirk Van den Poel
The paper presents a multi‑agent framework that uses fine‑tuned large language models with retrieval‑augmented generation to simulate partisan coalition negotiations and introduces…
InfoFlow KV: Information-Flow-Aware KV Recomputation for Long Context
Xin Teng, Canyu Zhang, Shaoyi Zheng +3
The paper proposes InfoFlow KV, a method that uses an attention‑norm signal to identify which key‑value cache tokens should be recomputed during retrieval‑augmented generation, imp…
Is External Database Protection Static in Retrieval-Augmented Generation? Rethinking Privacy Preservation under Dynamic Queries
Gang Zhang, Mingyu Tian, Xukun Luan +2
The paper studies privacy leaks in retrieval‑augmented generation and introduces a prompt‑aware dynamic hierarchical differential privacy method that adapts protection based on the…
SmartRAG: Native Graph-Based RAG for Mobile Device
Zhihan Jiang, Meng Li, Shenghao Liu +6
SmartRAG is an on-device framework that combines a small quantized language model with a graph-based retrieval system and a continually learnable named-entity recognizer to enable…
With Argus Eyes: Assessing Retrieval Gaps via Uncertainty Scoring to Detect and Remedy Retrieval Blind Spots
Zeinab Sadat Taghavi, Ali Modarressi, Hinrich Schutze +1
The paper identifies blind spots in neural retrievers used for retrieval‑augmented generation, where relevant entities are missed due to low embedding similarity, and proposes an u…
Not All Retrievals are Useful: Cross-Attention for Input-Aware RAG in Time Series Forecasting
Seunghan Lee, Jaehoon Lee, Jun Seo +7
The paper introduces Cross-RAG, a retrieval-augmented generation framework for zero-shot time series forecasting that uses query‑retrieval cross‑attention to selectively attend to…
When Reasoning Hurts: Source-Aware Evaluation of Frontier LLMs for Clinical SOAP Note Generation
Faizan Faisal
The paper evaluates whether reasoning abilities of large language models improve the generation of structured clinical SOAP notes, finding that reasoning can actually degrade perfo…
Learning Engagement Assistant (LEA): Cross-Course Scalability and Classroom Evaluation of an Agentic AI Tutoring System
Teri Rumble, Javad Zarrin, P. George Lovell +1
The paper reports on the real‑world classroom deployment of LEA, an adaptive AI tutoring assistant that combines retrieval‑augmented generation with structured knowledge component…
Earthquaker-AI: A Retrieval-Augmented Generation Framework with Rubric-Based Assessment for Primary School Earthquake Education
Xanthi Kokkinou, Chaido Mizeli, Nafsika Koulaxidou +2
Earthquaker-AI is a hybrid educational system that combines Lego robotics with a retrieval‑augmented generation AI assistant to teach primary‑school students earthquake safety, pro…