ArborMem: Navigating Interaction States with Memory Forests

📄 arXiv: 2608.17534v1 📥 PDF

作者: Zongwei Lv, Yuemeng Xu, Yilun Yao, Siyi Ding, Xinyu Tan, Yaoming Li, Guangxiang Zhao, Weihong Lin, Lin Sun, Xiangzheng Zhang, Tong Yang

分类: cs.CL

发布日期: 2026-08-18

备注: 24 pages, 2 figures


💡 一句话要点

提出ArborMem以解决对话状态记忆管理问题

🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 对话系统 记忆管理 交互状态 大型语言模型 在线记忆框架 多任务处理 上下文恢复

📋 核心要点

  1. 现有对话系统在处理多任务和多用户对话时,未能有效识别当前输入应恢复的先前对话状态,导致记忆管理不够灵活。
  2. ArborMem通过构建可导航的交互状态森林,能够在每次输入时定位相关状态,恢复局部上下文,并增强其与可重用证据的结合。
  3. 在多个基准测试中,ArborMem的表现超越了最强基线,尤其在受限读取预算下,其优势更加明显。

📝 摘要(中文)

随着大型语言模型越来越多地作为持久的对话助手,如何有效管理记忆以保持对话的连贯性成为一个重要问题。现有方法在处理对话历史时,往往未能明确当前输入应恢复的先前对话状态,尤其在多任务、多用户的对话场景中更为突出。本文提出了ArborMem,一个在线记忆框架,将长时间运行的对话表示为可导航的交互状态森林。每个分支保留局部一致的轨迹,而森林则维护多个可恢复的轨迹。实验结果表明,ArborMem在多个基准测试中显著优于现有方法,提升幅度在3.36到10.31个百分点之间。

🔬 方法详解

问题定义:本文旨在解决现有对话系统在多任务和多用户场景下,无法有效管理和恢复对话状态的问题。现有方法通常仅关注信息检索,而忽视了对话状态的上下文恢复。

核心思路:ArborMem的核心思想是将对话表示为一个可导航的交互状态森林,通过每个分支保留局部一致的轨迹,使得系统能够在对话中灵活地恢复和切换状态。

技术框架:ArborMem的整体架构包括状态定位、上下文恢复和证据增强三个主要模块。每当接收到新的输入时,系统首先定位相关的对话状态,然后恢复该状态的上下文,并从其他分支中检索可重用的证据。

关键创新:ArborMem的主要创新在于引入了交互状态森林的概念,使得对话系统能够在复杂的对话场景中有效管理多条轨迹,避免了语义相关但结构不同的轨迹混淆。

关键设计:在设计上,ArborMem采用了高效的状态定位算法,并结合了多种上下文恢复策略,确保在读取预算受限的情况下,系统的响应时间保持在半秒以下。

🖼️ 关键图片

fig_0
img_1
img_2

📊 实验亮点

在多个基准测试中,ArborMem的表现超越了最强基线,提升幅度在3.36到10.31个百分点之间,且在BranchMemEval上提升了5.0个百分点。其优势在受限读取预算下尤为明显,响应时间保持在半秒以下。

🎯 应用场景

ArborMem的研究成果可广泛应用于智能客服、虚拟助手和社交机器人等领域,提升其在复杂对话场景中的表现和用户体验。未来,该框架有望进一步推动对话系统的智能化和人性化发展。

📄 摘要(原文)

Large language models increasingly serve as persistent conversational assistants, requiring memory that preserves relevant experience and maintains continuity across interactions. Existing methods improve access to conversational history through long-context processing, selective retrieval, and structured memory organization. However, most systems treat memory access as retrieving relevant past information without first determining which prior interaction state the current turn resumes. This limitation becomes particularly important when conversations interleave multiple tasks, people, and plans that may be interrupted and later revisited. We introduce ArborMem, an online memory framework that represents a long-running conversation as a navigable forest of interaction states. Each branch preserves a locally coherent trajectory, while the forest maintains multiple trajectories that may later be resumed. For each new input, ArborMem localizes the relevant state, restores its branch-local context, and augments it with reusable evidence retrieved across branches, preserving interaction continuity without conflating semantically related but structurally distinct trajectories. Existing long-term memory benchmarks cover diverse memory and reasoning capabilities but do not explicitly isolate branch-structured challenges. We therefore introduce BranchMemEval, a controlled diagnostic benchmark for interleaved and resumable interaction trajectories. Experiments on LongMemEval, LoCoMo, BEAM 100K, and BranchMemEval show that ArborMem outperforms the strongest baselines by 3.36 to 10.31 percentage points on the three established benchmarks and by 5.0 points on BranchMemEval. Its advantage grows under constrained read budgets, while complete memory queries remain below half a second.