ParetoRAG: Leveraging Sentence-Context Attention for Robust and Efficient Retrieval-Augmented Generation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Yao, Ruobing, Zhang, Yifei, Song, Shuang, Liu, Yuhua, Gao, Neng, Tu, Chenyang
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912230420774912
author Yao, Ruobing
Zhang, Yifei
Song, Shuang
Liu, Yuhua
Gao, Neng
Tu, Chenyang
author_facet Yao, Ruobing
Zhang, Yifei
Song, Shuang
Liu, Yuhua
Gao, Neng
Tu, Chenyang
contents While Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) by incorporating external knowledge, they still face persistent challenges in retrieval inefficiency and the inability of LLMs to filter out irrelevant information. We present ParetoRAG, an unsupervised framework that optimizes RAG systems through sentence-level refinement guided by the Pareto principle. By decomposing paragraphs into sentences and dynamically re-weighting core content while preserving contextual coherence, ParetoRAG achieves dual improvements in both retrieval precision and generation quality without requiring additional training or API resources. This framework has been empirically validated across various datasets, LLMs, and retrievers.
format Preprint
id arxiv_https___arxiv_org_abs_2502_08178
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ParetoRAG: Leveraging Sentence-Context Attention for Robust and Efficient Retrieval-Augmented Generation
Yao, Ruobing
Zhang, Yifei
Song, Shuang
Liu, Yuhua
Gao, Neng
Tu, Chenyang
Computation and Language
While Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) by incorporating external knowledge, they still face persistent challenges in retrieval inefficiency and the inability of LLMs to filter out irrelevant information. We present ParetoRAG, an unsupervised framework that optimizes RAG systems through sentence-level refinement guided by the Pareto principle. By decomposing paragraphs into sentences and dynamically re-weighting core content while preserving contextual coherence, ParetoRAG achieves dual improvements in both retrieval precision and generation quality without requiring additional training or API resources. This framework has been empirically validated across various datasets, LLMs, and retrievers.
title ParetoRAG: Leveraging Sentence-Context Attention for Robust and Efficient Retrieval-Augmented Generation
topic Computation and Language
url https://arxiv.org/abs/2502.08178