@inproceedings{chen-etal-2026-beyond-semantic,
title = "Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents",
author = "Chen, Yaoqi and
Lai, Haibin and
Feng, Yuru and
Han, Chuyu and
Zhang, Qianxi and
Lu, Baotong and
Li, Menghao and
Wang, Xinjiang and
Jin, Zewen and
Wu, Hao and
Li, Cheng and
Chen, Qi",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.828/",
pages = "19255--19272",
ISBN = "979-8-89176-459-0",
abstract = "LLM-based agents increasingly tackle long-horizon tasks with interdependent decisions, where each action reshapes future constraints and intermediate errors can cascade. Existing RAG and agent memory systems organize histories by semantic similarity, retrieving content-relevant entries at decision time. We argue that this design mismatches execution-state dependencies: it fragments decision trajectories and mixes valid and erroneous traces, hindering coherent state reconstruction and error isolation. We propose MAGE (**M**emory as **A**gent-**G**uided **E**xploration), an active execution-state manager that stores interactions in a hierarchical state tree. The agent derives its state from the active root-to-current path, combining subgoal summaries, recent traces, and hints from prior branches. Four coupled operations maintain the tree: *Grow* records new traces, *Compress* summarizes completed subgoals, *Maintain* validates summaries, and *Revise* restores a target boundary and resumes on a new branch. This design bounds context growth while preserving state integrity and isolating flawed segments from the active path. Experiments on MemoryArena show that MAGE improves the average task success rate by 7.8{--}20.4 pp over baselines, while reducing token consumption by 55.1{\%}. Code is available at: https://github.com/Komorebi660/MAGE."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-beyond-semantic">
<titleInfo>
<title>Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yaoqi</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haibin</namePart>
<namePart type="family">Lai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuru</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chuyu</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qianxi</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Baotong</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Menghao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinjiang</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zewen</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cheng</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qi</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>LLM-based agents increasingly tackle long-horizon tasks with interdependent decisions, where each action reshapes future constraints and intermediate errors can cascade. Existing RAG and agent memory systems organize histories by semantic similarity, retrieving content-relevant entries at decision time. We argue that this design mismatches execution-state dependencies: it fragments decision trajectories and mixes valid and erroneous traces, hindering coherent state reconstruction and error isolation. We propose MAGE (**M**emory as **A**gent-**G**uided **E**xploration), an active execution-state manager that stores interactions in a hierarchical state tree. The agent derives its state from the active root-to-current path, combining subgoal summaries, recent traces, and hints from prior branches. Four coupled operations maintain the tree: *Grow* records new traces, *Compress* summarizes completed subgoals, *Maintain* validates summaries, and *Revise* restores a target boundary and resumes on a new branch. This design bounds context growth while preserving state integrity and isolating flawed segments from the active path. Experiments on MemoryArena show that MAGE improves the average task success rate by 7.8–20.4 pp over baselines, while reducing token consumption by 55.1%. Code is available at: https://github.com/Komorebi660/MAGE.</abstract>
<identifier type="citekey">chen-etal-2026-beyond-semantic</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.828/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19255</start>
<end>19272</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents
%A Chen, Yaoqi
%A Lai, Haibin
%A Feng, Yuru
%A Han, Chuyu
%A Zhang, Qianxi
%A Lu, Baotong
%A Li, Menghao
%A Wang, Xinjiang
%A Jin, Zewen
%A Wu, Hao
%A Li, Cheng
%A Chen, Qi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F chen-etal-2026-beyond-semantic
%X LLM-based agents increasingly tackle long-horizon tasks with interdependent decisions, where each action reshapes future constraints and intermediate errors can cascade. Existing RAG and agent memory systems organize histories by semantic similarity, retrieving content-relevant entries at decision time. We argue that this design mismatches execution-state dependencies: it fragments decision trajectories and mixes valid and erroneous traces, hindering coherent state reconstruction and error isolation. We propose MAGE (**M**emory as **A**gent-**G**uided **E**xploration), an active execution-state manager that stores interactions in a hierarchical state tree. The agent derives its state from the active root-to-current path, combining subgoal summaries, recent traces, and hints from prior branches. Four coupled operations maintain the tree: *Grow* records new traces, *Compress* summarizes completed subgoals, *Maintain* validates summaries, and *Revise* restores a target boundary and resumes on a new branch. This design bounds context growth while preserving state integrity and isolating flawed segments from the active path. Experiments on MemoryArena show that MAGE improves the average task success rate by 7.8–20.4 pp over baselines, while reducing token consumption by 55.1%. Code is available at: https://github.com/Komorebi660/MAGE.
%U https://aclanthology.org/2026.emnlp-main.828/
%P 19255-19272
Markdown (Informal)
[Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents](https://aclanthology.org/2026.emnlp-main.828/) (Chen et al., EMNLP 2026)
ACL
- Yaoqi Chen, Haibin Lai, Yuru Feng, Chuyu Han, Qianxi Zhang, Baotong Lu, Menghao Li, Xinjiang Wang, Zewen Jin, Hao Wu, Cheng Li, and Qi Chen. 2026. Beyond Semantic Organization: Memory as Execution State Management for Long-Horizon Agents. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 19255–19272, Budapest, Hungary. Association for Computational Linguistics.