@inproceedings{zhang-zeng-2026-tokens,
title = "From Tokens to Topology: When Large Language Models Meet Graph-Structured Reasoning",
author = "Zhang, Qihang and
Zeng, Liang",
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.1826/",
pages = "38990--39012",
ISBN = "979-8-89176-462-0",
abstract = "Reasoning benchmarks often collapse heterogeneous problem types into aggregate scores, obscuring which structural computations large language models (LLMs) can reliably perform. Existing graph-reasoning benchmarks cover graph traversal, broad graph computation, and graph-language interaction, but do not directly test whether graph-structured reasoning transfers across topological mechanisms under matched conditions. We introduce GraphReason-Bench (GR-Bench), comprising 1,800 programmatically generated and oracle-verified problems across three prompt-length regimes and six task families: local motif enumeration, exact-depth breadth-first search, bounded path enumeration, articulation-point identification, non-induced subgraph matching, and non-monotonic rule execution. All tasks share a text-only executable interface, exact oracles, canonical output normalization, and set-based scoring, enabling controlled cross-mechanism comparison. On the 600-instance main evaluation split across nine proprietary and open-source LLMs, we find mechanism-level dissociation: models that succeed on one structural operation often fail on others, and cross-task performance does not reduce to a stable graph-reasoning ability. Trace-level cases reveal mechanism-specific failures such as premature search termination and loss of intermediate state. An initial sparse-autoencoder probe, supported by cross-model robustness checks, shows that topology-sensitive representations become less separable as structural difficulty increases. Together, these results establish graph-structured reasoning as a measurable domain of LLM evaluation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-zeng-2026-tokens">
<titleInfo>
<title>From Tokens to Topology: When Large Language Models Meet Graph-Structured Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qihang</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liang</namePart>
<namePart type="family">Zeng</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>Reasoning benchmarks often collapse heterogeneous problem types into aggregate scores, obscuring which structural computations large language models (LLMs) can reliably perform. Existing graph-reasoning benchmarks cover graph traversal, broad graph computation, and graph-language interaction, but do not directly test whether graph-structured reasoning transfers across topological mechanisms under matched conditions. We introduce GraphReason-Bench (GR-Bench), comprising 1,800 programmatically generated and oracle-verified problems across three prompt-length regimes and six task families: local motif enumeration, exact-depth breadth-first search, bounded path enumeration, articulation-point identification, non-induced subgraph matching, and non-monotonic rule execution. All tasks share a text-only executable interface, exact oracles, canonical output normalization, and set-based scoring, enabling controlled cross-mechanism comparison. On the 600-instance main evaluation split across nine proprietary and open-source LLMs, we find mechanism-level dissociation: models that succeed on one structural operation often fail on others, and cross-task performance does not reduce to a stable graph-reasoning ability. Trace-level cases reveal mechanism-specific failures such as premature search termination and loss of intermediate state. An initial sparse-autoencoder probe, supported by cross-model robustness checks, shows that topology-sensitive representations become less separable as structural difficulty increases. Together, these results establish graph-structured reasoning as a measurable domain of LLM evaluation.</abstract>
<identifier type="citekey">zhang-zeng-2026-tokens</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1826/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>38990</start>
<end>39012</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Tokens to Topology: When Large Language Models Meet Graph-Structured Reasoning
%A Zhang, Qihang
%A Zeng, Liang
%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 zhang-zeng-2026-tokens
%X Reasoning benchmarks often collapse heterogeneous problem types into aggregate scores, obscuring which structural computations large language models (LLMs) can reliably perform. Existing graph-reasoning benchmarks cover graph traversal, broad graph computation, and graph-language interaction, but do not directly test whether graph-structured reasoning transfers across topological mechanisms under matched conditions. We introduce GraphReason-Bench (GR-Bench), comprising 1,800 programmatically generated and oracle-verified problems across three prompt-length regimes and six task families: local motif enumeration, exact-depth breadth-first search, bounded path enumeration, articulation-point identification, non-induced subgraph matching, and non-monotonic rule execution. All tasks share a text-only executable interface, exact oracles, canonical output normalization, and set-based scoring, enabling controlled cross-mechanism comparison. On the 600-instance main evaluation split across nine proprietary and open-source LLMs, we find mechanism-level dissociation: models that succeed on one structural operation often fail on others, and cross-task performance does not reduce to a stable graph-reasoning ability. Trace-level cases reveal mechanism-specific failures such as premature search termination and loss of intermediate state. An initial sparse-autoencoder probe, supported by cross-model robustness checks, shows that topology-sensitive representations become less separable as structural difficulty increases. Together, these results establish graph-structured reasoning as a measurable domain of LLM evaluation.
%U https://aclanthology.org/2026.findings-emnlp.1826/
%P 38990-39012
Markdown (Informal)
[From Tokens to Topology: When Large Language Models Meet Graph-Structured Reasoning](https://aclanthology.org/2026.findings-emnlp.1826/) (Zhang & Zeng, Findings 2026)
ACL