arXiv 2601.051712026-01-25cs.CL

Inside Out: 为长期个性化对话系统演化用户导向的核心内存树

Inside Out: Evolving User-Centric Core Memory Trees for Long-Term Personalized Dialogue Systems

School of Information, Renmin University of China（中国人民大学信息学院）；MemTensor (Shanghai) Technology Co., Ltd.（MemTensor（上海）科技有限公司）；Institute for Advanced Algorithms Research, Shanghai（上海先进算法研究所）；China Telecom Research Institute（中国电信研究院）；Beijing University of Aeronautics and Astronautics（北京航空航天大学）；Nankai University（南开大学）

Jihao Zhao, Ding Chen, Zhaoxin Fan, Kerun Xu, Mengting Hu, Bo Tang, Feiyu Xiong, Zhiyu Li

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

Inside Out通过演化用户导向的核心内存树，提升长期个性化对话系统的内存压缩与身份一致性，采用轻量级MemListener实现动态更新与高效响应生成。

摘要：

现有的长期个性化对话系统难以调和无界交互流与有限上下文约束，常陷入记忆噪声累积、推理退化和身份不一致的问题。为解决这些挑战，本文提出Inside Out框架，利用全局维护的PersonaTree作为长期用户画像的载体。通过约束树干以初始模式，并更新分支和叶子，PersonaTree实现了可控增长，在压缩内存的同时保持一致性。此外，我们通过基于进程的奖励训练轻量级MemListener，生成结构化、可执行且可解释的{ADD, UPDATE, DELETE, NO_OP}操作，从而支持个性化树的动态演化。在响应生成时，PersonaTree直接用于提升延迟敏感场景的输出；当用户需要更多细节时，代理模式被触发，在PersonaTree约束下按需引入细节。实验表明，PersonaTree在抑制上下文噪声和保持身份一致性方面优于全文拼接和各种个性化记忆系统。值得注意的是，小型MemListener模型在内存操作决策性能上可与甚至超越强大的推理模型如DeepSeek-R1-0528和Gemini-3-Pro。

英文摘要：

Existing long-term personalized dialogue systems struggle to reconcile unbounded interaction streams with finite context constraints, often succumbing to memory noise accumulation, reasoning degradation, and persona inconsistency. To address these challenges, this paper proposes Inside Out, a framework that utilizes a globally maintained PersonaTree as the carrier of long-term user profiling. By constraining the trunk with an initial schema and updating the branches and leaves, PersonaTree enables controllable growth, achieving memory compression while preserving consistency. Moreover, we train a lightweight MemListener via reinforcement learning with process-based rewards to produce structured, executable, and interpretable {ADD, UPDATE, DELETE, NO_OP} operations, thereby supporting the dynamic evolution of the personalized tree. During response generation, PersonaTree is directly leveraged to enhance outputs in latency-sensitive scenarios; when users require more details, the agentic mode is triggered to introduce details on-demand under the constraints of the PersonaTree. Experiments show that PersonaTree outperforms full-text concatenation and various personalized memory systems in suppressing contextual noise and maintaining persona consistency. Notably, the small MemListener model achieves memory-operation decision performance comparable to, or even surpassing, powerful reasoning models such as DeepSeek-R1-0528 and Gemini-3-Pro.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。