StructMem: Structured Memory for Long-Horizon Behavior in LLMs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Buqiang, Chen, Yijun, Fang, Jizhan, Zhong, Ruobin, Yao, Yunzhi, Zhu, Yuqi, Du, Lun, Deng, Shumin
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915952187146240
author Xu, Buqiang
Chen, Yijun
Fang, Jizhan
Zhong, Ruobin
Yao, Yunzhi
Zhu, Yuqi
Du, Lun
Deng, Shumin
author_facet Xu, Buqiang
Chen, Yijun
Fang, Jizhan
Zhong, Ruobin
Yao, Yunzhi
Zhu, Yuqi
Du, Lun
Deng, Shumin
contents Long-term conversational agents need memory systems that capture relationships between events, not merely isolated facts, to support temporal reasoning and multi-hop question answering. Current approaches face a fundamental trade-off: flat memory is efficient but fails to model relational structure, while graph-based memory enables structured reasoning at the cost of expensive and fragile construction. To address these issues, we propose \textbf{StructMem}, a structure-enriched hierarchical memory framework that preserves event-level bindings and induces cross-event connections. By temporally anchoring dual perspectives and performing periodic semantic consolidation, StructMem improves temporal reasoning and multi-hop performance on \texttt{LoCoMo}, while substantially reducing token usage, API calls, and runtime compared to prior memory systems, see https://github.com/zjunlp/LightMem .
format Preprint
id arxiv_https___arxiv_org_abs_2604_21748
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle StructMem: Structured Memory for Long-Horizon Behavior in LLMs
Xu, Buqiang
Chen, Yijun
Fang, Jizhan
Zhong, Ruobin
Yao, Yunzhi
Zhu, Yuqi
Du, Lun
Deng, Shumin
Computation and Language
Artificial Intelligence
Information Retrieval
Machine Learning
Multiagent Systems
Long-term conversational agents need memory systems that capture relationships between events, not merely isolated facts, to support temporal reasoning and multi-hop question answering. Current approaches face a fundamental trade-off: flat memory is efficient but fails to model relational structure, while graph-based memory enables structured reasoning at the cost of expensive and fragile construction. To address these issues, we propose \textbf{StructMem}, a structure-enriched hierarchical memory framework that preserves event-level bindings and induces cross-event connections. By temporally anchoring dual perspectives and performing periodic semantic consolidation, StructMem improves temporal reasoning and multi-hop performance on \texttt{LoCoMo}, while substantially reducing token usage, API calls, and runtime compared to prior memory systems, see https://github.com/zjunlp/LightMem .
title StructMem: Structured Memory for Long-Horizon Behavior in LLMs
topic Computation and Language
Artificial Intelligence
Information Retrieval
Machine Learning
Multiagent Systems
url https://arxiv.org/abs/2604.21748