Edit Knowledge, Not Just Facts via Multi-Step Reasoning over Background Stories

TL;DR

This paper proposes a reasoning-based knowledge editing method for large language models, using background stories and multi-step reasoning to better integrate new information into the model's understanding.

Contribution

It introduces a novel training strategy that emphasizes reasoning over memorization, involving background stories, multi-hop questions, and knowledge distillation.

Findings

Models effectively leverage new knowledge during reasoning.

Significant improvement on complex questions requiring multiple new facts.

Training strategy enhances knowledge internalization beyond factual recall.

Abstract

Enabling artificial intelligence systems, particularly large language models, to integrate new knowledge and flexibly apply it during reasoning remains a central challenge. Existing knowledge editing approaches emphasize atomic facts, improving factual recall but often failing to integrate new information into a coherent framework usable across contexts. In this work, we argue that knowledge internalization is fundamentally a reasoning problem rather than a memorization problem. Consequently, a model should be trained in situations where the new information is instrumental to solving a task, combined with pre-existing knowledge, and exercised through multi-step reasoning. Based on this insight, we propose a training strategy based on three principles. First, new knowledge is introduced as a coherent background story that contextualizes novel facts and explains their relation to existing knowledge. Second, models are trained using self-generated multi-hop questions that require multi-step reasoning involving the new information. Third, training is done using knowledge distillation, forcing a student model to internalize the teacher's reasoning behavior without access to the novel information. Experiments show that models trained with this strategy effectively leverage newly acquired knowledge during reasoning and achieve remarkable performance on challenging questions that require combining multiple new facts.