Improving Retrieval-Augmented Generation without Taxonomy-based Error Categorization
The paper presents a new approach to improving Retrieval-Augmented Generation (RAG) without relying on taxonomy-based error categorization. The proposed method, called RePAIR, focuses on directly mapping flawed outputs to corrective actions. This approach has shown consistent performance improvements across various benchmarks.
- ▪Retrieval-Augmented Generation enhances the accuracy of large language model outputs by integrating external knowledge.
- ▪The RePAIR paradigm eliminates the need for detailed error taxonomies and critic supervision.
- ▪Empirical results indicate that RePAIR significantly boosts the performance of agentic RAG systems.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Information Retrieval arXiv:2605.18772 (cs) [Submitted on 16 Apr 2026] Title:Improving Retrieval-Augmented Generation without Taxonomy-based Error Categorization Authors:Gongbo Zhang, Yifan Peng, Chunhua Weng View a PDF of the paper titled Improving Retrieval-Augmented Generation without Taxonomy-based Error Categorization, by Gongbo Zhang and 2 other authors View PDF HTML (experimental) Abstract:Retrieval-Augmented Generation (RAG) improves the factual accuracy of large language model (LLM) outputs by grounding generation in external knowledge. Recent agentic RAG systems extend this paradigm with critical agents to evaluate model responses and iteratively refine outputs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
