arXiv cs.CL· Junyou Guo, Chongyang Tao, Yan He, Yong Wang, Shuai Ma·· 6 天前AI 评分25
T-Mem：面向长程多方对话记忆的时间证据森林框架
Temporal Evidence Forests for Long-Horizon Multi-Party Conversational Memory
AI 导读
针对 LLM 智能体长期多方交互中时间结构难以利用的问题，研究者提出 T-Mem 记忆框架，通过时间证据发现与时间记忆结构化两项机制恢复局部密集证据并重建事件演进，再经证据森林线性化生成答案。在两个长期多方记忆基准上，T-Mem 在不同 LLM、参数设置、领域和查询类型下均持续优于现有竞争性记忆框架。
正文 · 原文
Abstract:Long-term memory has been widely adopted in LLM-based agents to support persistent interaction and personalized assistance. Long-term multi-party interactions, however, accumulate substantially more detailed information, with decisions repeatedly updated and states evolving across participants and interaction stages. Temporal structure is therefore crucial for tracing such memories: relevant information often appears densely within short intervals, while transitions among different states are anchored by a strict temporal order. Existing temporally aware memory frameworks primarily store time as metadata or use temporal signals to guide retrieval over preconstructed memory structures, without using temporal structure to both recover locally concentrated evidence and reconstruct its complete event progression. To address these limitations, we propose T-Mem, a temporal-structure-based memory framework that preserves interactions and organizes query-relevant evidence through two temporal mechanisms. Temporal Evidence Discovery exploits the local concentration of related messages to recover overlooked evidence, while Temporal Memory Structuring models state transitions and supporting relations over time and compresses low-information spans within the resulting forest. Evidence Forest Linearization then serializes the structured evidence for answer generation. Experiments on two long-term multi-party memory benchmarks demonstrate that T-Mem consistently outperforms competitive memory frameworks across LLMs, parameter settings, domains, and query types.
| Comments: | 32 pages, including appendices |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.12874 [cs.CL] |
| (or arXiv:2610.12874v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12874 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Junyou Guo [view email]
[v1]
Fri, 9 Oct 2026 02:05:38 UTC (5,012 KB)
来源：arXiv cs.CL · arxiv.org