通过诊断证据缺口，让对话记忆检索更准更快。
EviMem: Evidence-Gap-Driven Iterative Retrieval for Long-Term Conversational Memory
- 用闭环机制检测检索结果缺什么，指导精准改写问题
- 在多轮对话中，时间类和多跳问题准确率分别提升8.3%和19.3%
- 适合需要长期记忆的智能客服、对话系统研发者
长期对话记忆需从多轮会话中检索分散证据，但单次检索难以应对时序和多跳问题。现有迭代方法通过生成内容或文档级信号优化查询，但未明确诊断证据缺口（即当前检索集缺少什么），导致查询优化缺乏针对性。本文提出EviMem，结合IRIS（基于不足信号的迭代检索）与LaceMem（分层对话证据记忆架构），构建闭环框架：通过充分性评估检测证据缺口，诊断缺失内容，并驱动精准查询改写。在LoCoMo数据集上，相比MIRIX，EviMem在时间类问题上准确率从73.3%提升至81.6%，多跳问题从65.9%提升至85.2%，且延迟降低4.5倍。代码已开源。
原文摘要 · Abstract (English)
Long-term conversational memory requires retrieving evidence scattered across multiple sessions, yet single-pass retrieval fails on temporal and multi-hop questions. Existing iterative methods refine queries via generated content or document-level signals, but none explicitly diagnoses the evidence gap, namely what is missing from the accumulated retrieval set, leaving query refinement untargeted. We present EviMem, combining IRIS (Iterative Retrieval via Insufficiency Signals), a closed-loop framework that detects evidence gaps through sufficiency evaluation, diagnoses what is missing, and drives targeted query refinement, with LaceMem (Layered Architecture for Conversational Evidence Memory), a coarse-to-fine memory hierarchy supporting fine-grained gap diagnosis. On LoCoMo, EviMem improves Judge Accuracy over MIRIX on temporal (73.3% to 81.6%) and multi-hop (65.9% to 85.2%) questions at 4.5x lower latency. Code: https://github.com/AIGeeksGroup/EviMem.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。