Hierarchical Document Refinement for Long-context Retrieval-augmented Generation

TL;DR

This paper introduces LongRefiner, a hierarchical document refinement method that improves long-context retrieval-augmented generation by reducing computational costs and noise, while maintaining high performance across multiple QA datasets.

Contribution

LongRefiner is a novel, efficient plug-and-play refiner that leverages document structure and multi-task learning to enhance long-text RAG performance with significantly lower costs.

Findings

Achieves competitive accuracy on seven QA datasets.

Uses 10x less computational resources and latency.

Scalable and effective for real-world applications.

Abstract

Real-world RAG applications often encounter long-context input scenarios, where redundant information and noise results in higher inference costs and reduced performance. To address these challenges, we propose LongRefiner, an efficient plug-and-play refiner that leverages the inherent structural characteristics of long documents. LongRefiner employs dual-level query analysis, hierarchical document structuring, and adaptive refinement through multi-task learning on a single foundation model. Experiments on seven QA datasets demonstrate that LongRefiner achieves competitive performance in various scenarios while using 10x fewer computational costs and latency compared to the best baseline. Further analysis validates that LongRefiner is scalable, efficient, and effective, providing practical insights for real-world long-text RAG applications. Our code is available at https://github.com/ignorejjj/LongRefiner.