rEDMRec: Distilling Large Language Model Reasoning into an Editable Experience Memory for Recommendation
作者: Minh Hoang Nguyen, Tung Le, Huy Tien Nguyen
分类: cs.IR, cs.AI, cs.CL
发布日期: 2026-08-19
💡 一句话要点
提出rEDMRec以解决推荐系统中的推理效率问题
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 推荐系统 大型语言模型 推理效率 经验记忆 个性化推荐 机器学习 模型蒸馏
📋 核心要点
- 现有推荐系统在每次请求中重复推理用户偏好,导致计算成本高且无法重用推理结果。
- rEDMRec通过将推理结果压缩为结构化的经验记忆,利用轻量级模型进行信息检索,避免重复推理。
- 在多个数据集上,rEDMRec在不同模型上均提升了推荐准确率,尤其在ML-1M上提升幅度达到13.3%。
📝 摘要(中文)
大型语言模型能够通过明确推理用户历史和候选项目来提高推荐质量。然而,这种推理在每次排名请求中重复执行成本高昂,且通常一次性消费后被丢弃,导致无法在未来请求中重用。为此,本文提出rEDMRec,通过将教师模型的推理压缩为四个可编辑的经验通道,利用轻量级模型从中检索信息,从而降低在线推理成本。实验结果表明,rEDMRec在多个数据集上均显著提升了推荐效果,尤其在ML-1M数据集上,提升幅度达到13.3%。
🔬 方法详解
问题定义:本文旨在解决推荐系统中推理效率低下的问题。现有方法在每次请求中都需重复推理用户偏好,导致计算成本高且推理结果无法重用。
核心思路:rEDMRec的核心思想是将教师模型的推理结果压缩为四个可编辑的经验通道,允许轻量级模型从中检索信息,从而降低在线推理的成本。
技术框架:rEDMRec的整体架构包括一个LLM记忆控制器和一个轻量级学生模型。记忆控制器管理四个经验通道,执行添加、删除、修改和保留操作,而学生模型则从记忆中检索信息进行候选项排名。
关键创新:rEDMRec的主要创新在于将推理结果结构化为可编辑的经验通道,允许在不调用教师模型的情况下进行高效的推荐。这一设计显著降低了在线推理的计算负担。
关键设计:在设计中,经验通道包括长期偏好、短期上下文、项目感知和反事实对比。通过K-agent辩论优化记忆,减少了记忆重复率,并提升了下游推荐的准确性。
🖼️ 关键图片
📊 实验亮点
在多个数据集上,rEDMRec显著提升了推荐效果,尤其在ML-1M数据集上,相较于第二最佳基线,提升幅度达到13.3%。此外,短期上下文通道在不同模型中表现一致,而其他通道的贡献则依赖于模型的容量。
🎯 应用场景
rEDMRec的研究成果在个性化推荐系统中具有广泛的应用潜力,能够有效提升用户体验和推荐准确性。其结构化的经验记忆设计可为未来的推荐算法提供新的思路,尤其在用户偏好变化频繁的场景中,能够快速适应并优化推荐结果。
📄 摘要(原文)
Large language models can improve recommendation quality by reasoning explicitly over user history and candidate items - for example, extracting a user's preferences or explaining why one item fits better than another - rather than mapping history directly to a ranked list. This reasoning, however, is expensive to repeat on every ranking request and, once produced, is typically consumed once and discarded, leaving it neither reusable across future requests nor easy to inspect or correct as user tastes drift. Our insight is that reasoning does not need to be regenerated at every call if it can instead be compressed once into a compact, structured memory that a lightweight model retrieves from. We propose rEDMRec, which distills a teacher LLM's reasoning into four typed, editable experience channels - long-term preference, short-term context, item-perception, and counterfactual hard-negative comparisons - maintained by an LLM memory controller that performs Add/Delete/Modify/Keep operations and refines entries via K-agent debate. A lightweight student LLM then ranks candidates purely by retrieving from this memory, without invoking the teacher again, decoupling online inference cost from reasoning depth. Across ML-1M, Amazon Beauty, and Steam and ten student backbones, rEDMRec improves HR@1 over zero-shot, few-shot, and RAG on every backbone, and over GraphRAG on most backbones, with Impv up to 13.3% vs. the second-best baseline on ML-1M. Channel ablations show that short-term context is the only channel that helps consistently across capacity tiers, whereas long-term, item-perception, and counterfactual contributions are capacity-dependent (and can reverse on the strongest students); debate-based memory optimization lowers bank duplication by 7.4 percentage points while raising downstream HR@1 by up to +0.029 over six optimization epochs.