通过优化检索与生成路径，显著减少大模型幻觉并提升复杂推理准确率。
Self-Correcting RAG: Enhancing Faithfulness via MMKP Context Selection and NLI-Guided MCTS
- 将上下文选择建模为多维多重背包问题，高效利用有限token预算
- 采用NLI引导的MCTS动态探索推理路径，验证生成内容真实性
- 在6个数据集上验证效果，复杂问题准确率提升且幻觉大幅减少
检索增强生成（RAG）显著扩展了大语言模型的知识边界，但在处理复杂推理任务时仍面临上下文利用率低和频繁幻觉两大挑战。为此，我们提出Self-Correcting RAG，将检索与生成重构为约束优化与路径规划问题。输入端，突破传统贪心检索，首次将上下文选择形式化为多维多重背包问题（MMKP），在严格token预算下最大化信息密度并消除冗余；输出端，引入自然语言推理（NLI）引导的蒙特卡洛树搜索（MCTS）机制，利用推理时计算资源动态探索推理路径，并验证生成答案的忠实性。在六个多跳问答与事实核查数据集上的实验表明，该方法在复杂查询上显著提升推理准确率，同时有效降低幻觉，优于现有强基线。代码已开源。
原文摘要 · Abstract (English)
Retrieval-augmented generation (RAG) substantially extends the knowledge boundary of large language models. However, it still faces two major challenges when handling complex reasoning tasks: low context utilization and frequent hallucinations. To address these issues, we propose Self-Correcting RAG, a unified framework that reformulates retrieval and generation as constrained optimization and path planning. On the input side, we move beyond traditional greedy retrieval and, for the first time, formalize context selection as a multi-dimensional multiple-choice knapsack problem (MMKP), thereby maximizing information density and removing redundancy under a strict token budget. On the output side, we introduce a natural language inference (NLI)-guided Monte Carlo Tree Search (MCTS) mechanism, which leverages test-time compute to dynamically explore reasoning trajectories and validate the faithfulness of generated answers. Experiments on six multi-hop question answering and fact-checking datasets demonstrate that our method significantly improves reasoning accuracy on complex queries while effectively reducing hallucinations, outperforming strong existing baselines.Our code is available at https://github.com/xjiacs/Self-Correcting-RAG .
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。