Robust Knowledge Editing via Explicit Reasoning Chains for Distractor-Resilient Multi-Hop QA
This addresses the challenge of timely knowledge integration in LLMs for applications requiring reliable updates, though it is incremental as it builds on existing knowledge-editing techniques.
The paper tackles the problem of updating knowledge in large language models (LLMs) without retraining, specifically in noisy, multi-hop question-answering (QA) scenarios, and achieves a multi-hop QA accuracy of 90.2% with only a 6.3% drop under heavy distraction.
Large language models (LLMs) encode vast amounts of world knowledge but remain static once trained, making the timely integration of emerging facts prohibitively expensive via full retraining. Knowledge-editing techniques have thus emerged to inject or overwrite specific facts into LLMs, yet they either over-rely on superficial cues or incur complex, iterative pipelines that collapse under noisy, multi-hop conditions. We introduce Reason-KE, an end-to-end reasoning-chain-based editing framework that steers a pretrained LLM through four structured stages-fact acknowledgment, relevance determination, selective application, and final reasoning-to filter distractors in a single pass. Trained on MQuAKE-CF with up to four irrelevant facts, Reason-KE elevates Qwen2.5-7B's multi-hop QA accuracy to 90.2% while suffering merely a 6.3% drop under heavy distraction and <1% when answers are leaked. Our quantitative analysis confirms Reason-KE's resilience and efficiency, establishing a new state-of-the-art for reliable LLM knowledge updates.