@inproceedings{jiang-etal-2026-evidence,
title = "When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems",
author = "Jiang, Yangxiao and
Fan, Jiarun and
Xu, Mingcong and
Guo, Yanxi and
Feng, Jiwen and
Xu, Shanqing and
Qian, Mengchen and
Chen, Wei and
Zhang, Xiaojin",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1219/",
pages = "25814--25829",
ISBN = "979-8-89176-462-0",
abstract = "Multi-Agent Systems (MAS) have recently moved from static workflows toward dynamically generated collaboration topologies. However, existing topology generation methods rely primarily on the parametric knowledge of large language models, with external search or retrieval used only as a reactive tool rather than an explicit determinant of collaboration structure. This leads to structure{--}knowledge misalignment, where systems exhibit redundant interactions or insufficient verification in knowledge-intensive tasks.\par We propose K-GAT (Knowledge-Guided Agent Topology Generator), a neuro-symbolic framework that formulates collaboration topology design as a knowledge-conditioned structure learning problem, integrating external evidence directly into autoregressive graph generation. Extensive experiments on knowledge-intensive benchmarks demonstrate K-GAT{'}s efficiency and effectiveness. Notably, on the expert-level GPQA dataset, K-GAT outperforms the LLM-Debate baseline by a substantial margin of +15.7{\%} in accuracy, while consuming less than half the computational tokens."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jiang-etal-2026-evidence">
<titleInfo>
<title>When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yangxiao</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiarun</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingcong</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanxi</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiwen</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shanqing</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mengchen</namePart>
<namePart type="family">Qian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaojin</namePart>
<namePart type="family">Zhang</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Multi-Agent Systems (MAS) have recently moved from static workflows toward dynamically generated collaboration topologies. However, existing topology generation methods rely primarily on the parametric knowledge of large language models, with external search or retrieval used only as a reactive tool rather than an explicit determinant of collaboration structure. This leads to structure–knowledge misalignment, where systems exhibit redundant interactions or insufficient verification in knowledge-intensive tasks.\par We propose K-GAT (Knowledge-Guided Agent Topology Generator), a neuro-symbolic framework that formulates collaboration topology design as a knowledge-conditioned structure learning problem, integrating external evidence directly into autoregressive graph generation. Extensive experiments on knowledge-intensive benchmarks demonstrate K-GAT’s efficiency and effectiveness. Notably, on the expert-level GPQA dataset, K-GAT outperforms the LLM-Debate baseline by a substantial margin of +15.7% in accuracy, while consuming less than half the computational tokens.</abstract>
<identifier type="citekey">jiang-etal-2026-evidence</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1219/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>25814</start>
<end>25829</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems
%A Jiang, Yangxiao
%A Fan, Jiarun
%A Xu, Mingcong
%A Guo, Yanxi
%A Feng, Jiwen
%A Xu, Shanqing
%A Qian, Mengchen
%A Chen, Wei
%A Zhang, Xiaojin
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F jiang-etal-2026-evidence
%X Multi-Agent Systems (MAS) have recently moved from static workflows toward dynamically generated collaboration topologies. However, existing topology generation methods rely primarily on the parametric knowledge of large language models, with external search or retrieval used only as a reactive tool rather than an explicit determinant of collaboration structure. This leads to structure–knowledge misalignment, where systems exhibit redundant interactions or insufficient verification in knowledge-intensive tasks.\par We propose K-GAT (Knowledge-Guided Agent Topology Generator), a neuro-symbolic framework that formulates collaboration topology design as a knowledge-conditioned structure learning problem, integrating external evidence directly into autoregressive graph generation. Extensive experiments on knowledge-intensive benchmarks demonstrate K-GAT’s efficiency and effectiveness. Notably, on the expert-level GPQA dataset, K-GAT outperforms the LLM-Debate baseline by a substantial margin of +15.7% in accuracy, while consuming less than half the computational tokens.
%U https://aclanthology.org/2026.findings-emnlp.1219/
%P 25814-25829
Markdown (Informal)
[When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems](https://aclanthology.org/2026.findings-emnlp.1219/) (Jiang et al., Findings 2026)
ACL
- Yangxiao Jiang, Jiarun Fan, Mingcong Xu, Yanxi Guo, Jiwen Feng, Shanqing Xu, Mengchen Qian, Wei Chen, and Xiaojin Zhang. 2026. When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 25814–25829, Budapest, Hungary. Association for Computational Linguistics.