AdaGATE: Adaptive Gap-Aware Token-Efficient Evidence Assembly for Multi-Hop Retrieval-Augmented Generation

AdaGATE：用于多跳检索增强生成的自适应缺口感知与高效令牌证据组装

Abstract: Retrieval-augmented generation (RAG) remains brittle on multi-hop questions in realistic deployment settings, where retrieved evidence may be noisy or redundant and only limited context can be passed to the generator. Existing controllers address parts of this problem, but typically either expand context additively, select from a fixed top-k set, or optimize relevance without explicitly repairing missing bridge facts.

摘要： 在实际部署环境中，检索增强生成（RAG）在处理多跳问题时仍然表现脆弱，因为检索到的证据可能存在噪声或冗余，且传递给生成器的上下文极其有限。现有的控制器虽然解决了部分问题，但通常要么通过累加方式扩展上下文，要么从固定的 top-k 集合中进行选择，亦或是仅优化相关性，而无法显式地修复缺失的桥接事实。

We propose AdaGATE, a training-free evidence controller for multi-hop RAG that frames evidence selection as a token-constrained repair problem. AdaGATE combines entity centric gap tracking, targeted micro-query generation, and a utility based selection mechanism that balances gap coverage, corroboration, novelty, redundancy, and direct question relevance.

我们提出了 AdaGATE，这是一种用于多跳 RAG 的免训练证据控制器，它将证据选择建模为一个受令牌约束的修复问题。AdaGATE 结合了以实体为中心的缺口追踪、针对性的微查询生成，以及一种基于效用的选择机制，旨在平衡缺口覆盖率、证据佐证性、新颖性、冗余度以及与问题的直接相关性。

We evaluate AdaGATE on HotpotQA under clean, redundancy, and noise injected retrieval conditions. Across all three settings, AdaGATE achieves the best evidence F1 among the compared controllers, reaching 62.3% on clean data and 71.2% under redundancy injection, while using 2.6x fewer input tokens than Adaptive-k. These results suggest that explicit gap-aware repair, combined with token-efficient evidence selection, improves robustness in multi-hop RAG under imperfect retrieval.

我们在干净、冗余和注入噪声的检索条件下，对 HotpotQA 数据集进行了评估。在所有三种设置下，AdaGATE 在对比控制器中均取得了最佳的证据 F1 分数，在干净数据上达到 62.3%，在冗余注入条件下达到 71.2%，同时其使用的输入令牌数量比 Adaptive-k 少 2.6 倍。这些结果表明，显式的缺口感知修复与高效的令牌证据选择相结合，能够提升多跳 RAG 在不完美检索条件下的鲁棒性。

Our code and evaluation pipeline are available at this https URL.

我们的代码和评估流程已在以下链接提供：[https URL]。