Replace, Dont Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly

📝 Original Info

• Title: Replace, Dont Expand: Mitigating Context Dilution in Multi-Hop RAG via Fixed-Budget Evidence Assembly
• ArXiv ID: 2512.10787
• Date: 2025-12-11
• Authors: Moshe Lahmy, Roi Yozevitch

📝 Abstract

Retrieval-Augmented Generation (RAG) systems often fail on multi-hop queries when the initial retrieval misses a bridge fact. Prior corrective approaches, such as Self-RAG, CRAG, and Adaptivek, typically address this by adding more context or pruning existing lists. However, simply expanding the context window often leads to context dilution, where distractors crowd out relevant information. We propose SEAL-RAG, a training-free controller that adopts a "replace, don't expand" strategy to fight context dilution under a fixed retrieval depth k. SEAL executes a (Search → Extract → Assess → Loop) cycle: it performs on-the-fly, entity-anchored extraction to build a live gap specification (missing entities/relations), triggers targeted micro-queries, and uses entity-first ranking to actively swap out distractors for gap-closing evidence. We evaluate SEAL-RAG against faithful re-implementations of Basic RAG, CRAG, Self-RAG, and Adaptive-k in a shared environment on HotpotQA and 2WikiMultiHopQA. On HotpotQA (k = 3), SEAL improves answer correctness by +3 to +13 pp and evidence precision by +12 to +18 pp over Self-RAG. On 2Wiki-MultiHopQA (k = 5), it outperforms Adaptive-k by +8.0 pp in accuracy and maintains 96% evidence precision, ...

📄 Full Content

...(본문 내용이 길어 생략되었습니다. 사이트에서 전문을 확인해 주세요.)