FedMosaic: Federated Retrieval-Augmented Generation via Parametric Adapters
sigir26-131 · Full research paper · Zhilin Liang, Yuxiang Wang, Zimu Zhou, Hainan Zhang, Boyi Liu, Yongxin Tong
Efficient and Long-Context RAG · 100%
Abstract
Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by grounding generation in external knowledge to improve factuality and reduce hallucinations. Yet most deployments assume a centralized corpus, which is infeasible in privacy-aware domains where knowledge remains siloed. This motivates federated RAG (FedRAG), where a central LLM server collaborates with distributed silos without sharing raw documents. In-context RAG violates this requirement by transmitting verbatim documents, whereas parametric RAG encodes documents into light weight adapters that merge with a frozen LLM at inference, avoiding raw-text exchange. We adopt the parametric approach but face two unique challenges induced by FedRAG: high storage and communication from per-document adapters, and destructive aggregation caused by indiscriminately merging multiple adapters. We present FedMosaic, the first federated RAG framework built on parametric adapters. FedMosaic clusters semantically related documents into multi-document adapters with document-specific masks to reduce overhead while preserving specificity, and performs selective adapter aggregation to combine only relevance-aligned, non-conflicting adapters. Experiments show that FedMosaic achieves an average 10.9% higher accuracy than state-of-the-art methods in four categories, while lowering storage costs by 78.8% to 86.3% and communication costs by 91.4%, and never sharing raw documents.
Assigned reviewers
| # | Reviewer | Match | Why (closest evidence publications in the taxonomy) | |
|---|---|---|---|---|
| 1 | Peitian Zhang | 50 | 100 RetroLM: Retrieval-Augmented KVs for Long-Context Processing (2026) Efficient and Long-Context RAG 100 MemoRAG: Boosting Long Context Processing with Global Memory-Enhanced Retrieval Augmentation (2025) Efficient and Long-Context RAG 100 Boosting Long-Context Information Seeking via Query-Guided Activation Refilling (2025) Efficient and Long-Context RAG |
|
| 2 | Amir Ingber | 25 | 37 An Approximate Algorithm for Maximum Inner Product Search over Streaming Sparse Vectors (2024) Approximate Nearest Neighbor Search 37 Results of the Big ANN: NeurIPS'23 competition (2025) Approximate Nearest Neighbor Search 34 Bridging Dense and Sparse Maximum Inner Product Search (2024) Approximate Nearest Neighbor Search |
Area chair
| Area chair | Match | |
|---|---|---|
| Hongjin Qian | 39 |
Area-chair candidates (6)
| Hongjin Qian assigned | 39 | 12 papers | |
| Xiaoxi Li 0005 | 24 | 15 papers | |
| Pengyue Jia | 23 | 10 papers | |
| Fernando Diaz 0001 | 22 | 12 papers | |
| Maolin Wang 0001 | 22 | 15 papers | |
| Chenyan Xiong | 22 | 15 papers |
Candidates from the panel ranked by match score
| # | Reviewer | Match | Load | Why | |
|---|---|---|---|---|---|
| 1 | Peitian Zhang assigned | 50 | 3 | 100 RetroLM: Retrieval-Augmented KVs for Long-Context Processing (2026) Efficient and Long-Context RAG 100 MemoRAG: Boosting Long Context Processing with Global Memory-Enhanced Retrieval Augmentation (2025) Efficient and Long-Context RAG |
|
| 2 | Shu Zhou 0002 | 42 | 3 | 100 Latent Retrieval Augmented Generation (2026) Efficient and Long-Context RAG 100 Inference Scaling Law for Retrieval Augmented Generation (2026) Efficient and Long-Context RAG |
|
| 4 | Jingtao Zhan | 30 | 3 | 63 Dynamic and Parametric Retrieval-Augmented Generation (2025) Multi-Hop and Iterative Retrieval Reasoning 34 Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval (2022) Approximate Nearest Neighbor Search |
|
| 5 | Antonio Mallia | 30 | 3 | 32 Learning Passage Impacts for Inverted Indexes (2021) Learned Sparse Retrieval Efficiency 32 Improved Learned Sparse Retrieval with Corpus-Specific Vocabularies (2024) Learned Sparse Retrieval Efficiency |
|
| 6 | Franco Maria Nardini | 28 | 3 | 31 ReNeuIR at SIGIR 2026: The Fifth Workshop on Reaching Efficiency in Neural Information Retrieval (2026) Learned Sparse Retrieval Efficiency 31 ReNeuIR at SIGIR 2023: The Second Workshop on Reaching Efficiency in Neural Information Retrieval (2023) Learned Sparse Retrieval Efficiency |
|
| 7 | Di Wu 0054 | 28 | 3 | 64 Repoformer: Selective Retrieval for Repository-Level Code Completion (2024) Efficient and Long-Context RAG 27 LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (2025) Conversational Search and Retrieval |
|
| 8 | Huiqiang Jiang | 27 | 3 | 31 LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models (2023) Learned Sparse Retrieval Efficiency 31 SCBench: A KV Cache-Centric Analysis of Long-Context Methods (2025) Learned Sparse Retrieval Efficiency |
|
| 9 | Xinyu Ma 0001 | 27 | 3 | 66 Probe-and-Fetch: Dynamic KV Cache Pruning for Accelerated Long-Context Inference in Web-Scale AI Search (2026) Efficient and Long-Context RAG 28 PROP: Pre-training with Representative Words Prediction for Ad-hoc Retrieval (2021) Sparse and Dense Retrieval Encoders |
|
| 10 | Xueguang Ma | 27 | 3 | 32 LACONIC: Dense-Level Effectiveness for Scalable Sparse Retrieval via a Two-Phase Training Curriculum (2026) Learned Sparse Retrieval Efficiency 31 SLIM: Sparsified Late Interaction for Multi-Vector Retrieval with Inverted Indexes (2023) Learned Sparse Retrieval Efficiency |
|
| 11 | Jing Liu 0022 | 26 | 3 | 100 BEE-RAG: Balanced Entropy Engineering for Retrieval-Augmented Generation (2026) Efficient and Long-Context RAG 27 Dense Text Retrieval Based on Pretrained Language Models: A Survey (2024) Dense Bi-Encoder Training and Representation Learning |
|
| 12 | Yuchen Li 0006 | 25 | 3 | 66 Probe-and-Fetch: Dynamic KV Cache Pruning for Accelerated Long-Context Inference in Web-Scale AI Search (2026) Efficient and Long-Context RAG 27 When Search Engine Services Meet Large Language Models: Visions and Challenges (2024) Retrieval-Augmented Generation › Others |
|
| 13 | Sebastian Bruch 0001 | 25 | 3 | 34 Pairing Clustered Inverted Indexes with κ-NN Graphs for Fast Approximate Retrieval over Learned Sparse Representations (2024) Approximate Nearest Neighbor Search 32 Bridging Dense and Sparse Maximum Inner Product Search (2024) Efficient Retrieval and Indexing |
|
| 14 | Gyuseok Lee | 25 | 3 | 29 Continual Collaborative Distillation for Recommender System (2024) Efficient and Lightweight Graph Collaborative Filtering 24 Capturing User Interests from Data Streams for Continual Sequential Recommendation (2026) Efficient Sequential Recommendation Architectures |
|
| 15 | Amir Ingber assigned | 25 | 3 | 37 An Approximate Algorithm for Maximum Inner Product Search over Streaming Sparse Vectors (2024) Approximate Nearest Neighbor Search 37 Results of the Big ANN: NeurIPS'23 competition (2025) Approximate Nearest Neighbor Search |
|
| 16 | Sheng-Chieh Lin | 25 | 3 | 32 Efficiently Teaching an Effective Dense Retriever with Balanced Topic Aware Sampling (2021) Approximate Nearest Neighbor Search 31 SLIM: Sparsified Late Interaction for Multi-Vector Retrieval with Inverted Indexes (2023) Learned Sparse Retrieval Efficiency |
|
| 17 | Xiao Wang 0043 | 25 | 3 | 29 Reproducibility, Replicability, and Insights into Dense Multi-Representation Retrieval Models: from ColBERT to Col (2023) Learned Sparse Retrieval Efficiency 29 An Inspection of the Reproducibility and Replicability of TCT-ColBERT (2022) Learned Sparse Retrieval Efficiency |
|
| 18 | Yang Zhang 0072 | 24 | 3 | 29 Preliminary Study on Incremental Learning for Large Language Model-based Recommender Systems (2024) Efficiency and Deployment 28 Don't Start Over: A Cost-Effective Framework for Migrating Personalized Prompts Between LLMs (2026) Efficiency and Deployment |
|
| 19 | Xiangyu Ke COI: co-author | 24 | 3 | 37 Scalable Graph Indexing using GPUs for Approximate Nearest Neighbor Search (2025) Approximate Nearest Neighbor Search 37 Stitching Inner Product and Euclidean Metrics for Topology-aware Maximum Inner Product Search (2025) Approximate Nearest Neighbor Search |
|
| 21 | Wenlin Zhang 0001 | 24 | 3 | 29 A Survey of Personalization: From RAG to Agent (2026) Personalized and Conversational RAG 29 Bridging Personalization and AI: From RAG to Agent (2026) Personalized and Conversational RAG |
|
| 22 | Hung Vinh Tran | 24 | 3 | 29 A Thorough Performance Benchmarking on Lightweight Embedding-based Recommender Systems (2025) Recommendation Systems and Evaluation 25 On-device Content-based Recommendation with Single-shot Embedding Pruning: A Cooperative Game Perspective (2025) Content-Based Transferable Recommendation |
|
| 23 | Yujia Zhou 0002 | 24 | 3 | 100 Parametric Retrieval Augmented Generation (2025) Efficient and Long-Context RAG 25 Robust Fine-tuning for Retrieval Augmented Generation against Retrieval Defects (2025) Knowledge Conflict and Faithfulness |
|
| 24 | Shengyao Zhuang | 23 | 3 | 31 LACONIC: Dense-Level Effectiveness for Scalable Sparse Retrieval via a Two-Phase Training Curriculum (2026) Learned Sparse Retrieval Efficiency 29 Layer-wise Token Compression for Efficient Document Reranking (2026) Learned Sparse Retrieval Efficiency |
|
| 25 | Zheng Zhang 0006 | 23 | 3 | 23 Enhancing Cross-Modal Retrieval via Visual-Textual Prompt Hashing (2024) Cross-Modal Hashing and Quantization 23 Semantics-Reconstructing Hashing for Cross-Modal Retrieval (2020) Cross-Modal Hashing and Quantization |
|
| 26 | Weizhi Xu 0002 | 23 | 3 | 27 Graph-based Hierarchical Relevance Matching Signals for Ad-hoc Retrieval (2021) Dense Bi-Encoder Training and Representation Learning 23 Heterogeneous Graph Reasoning for Fact Checking over Texts and Tables (2024) Automated Fact Checking and Claim Verification |
|
| 27 | Zhe Xu 0009 | 23 | 3 | 25 Dynamic Demonstration Retrieval and Cognitive Understanding for Emotional Support Conversation (2024) Conversational Search and Retrieval 25 Intent-driven In-context Learning for Few-shot Dialogue State Tracking (2025) Conversational Search and Retrieval |
|
| 28 | Yingyi Zhang 0001 | 23 | 3 | 31 A Survey of Personalization: From RAG to Agent (2026) Personalized and Conversational RAG 31 Bridging Personalization and AI: From RAG to Agent (2026) Personalized and Conversational RAG |
|
| 29 | Venktesh V | 23 | 3 | 29 Breaking the Lens of the Telescope: Online Relevance Estimation over Large Retrieval Sets (2025) Learned Sparse Retrieval Efficiency 24 Reproducing Adaptive Reranking for Reasoning-Intensive IR (2026) Evidence Selection and Context Refinement |
|
| 31 | Corby Rosset | 22 | 3 | 31 Optimizing Query Evaluations Using Reinforcement Learning for Web Search (2018) Efficient Retrieval and Indexing 28 Augmenting Zero-Shot Dense Retrievers with Plug-in Mixture-of-Memories (2023) Retrieval-Augmented Generation › Others |
|
| 32 | Jianfeng Dong | 22 | 3 | 25 W2VV++: Fully Deep Learning for Ad-hoc Video Search (2019) Video and Moment Retrieval 25 Dual Learning with Dynamic Knowledge Distillation for Partially Relevant Video Retrieval (2023) Video and Moment Retrieval |
|
| 33 | Qisheng Hu | 22 | 3 | 23 STARE at the Structure: Steering ICL Exemplar Selection with Structural Alignment (2025) Language Model Pre-training and Adaptation 23 Decomposition Dilemmas: Does Claim Decomposition Boost or Burden Fact-Checking Performance? (2025) Automated Fact Checking and Claim Verification |