@inproceedings{lee-etal-2025-token,
title = "From Token to Action: State Machine Reasoning to Mitigate Overthinking in Information Retrieval",
author = "Lee, Dohyeon and
Jeong, Yeonseok and
Hwang, Seung-won",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.371/",
doi = "10.18653/v1/2025.findings-emnlp.371",
pages = "7048--7064",
ISBN = "979-8-89176-335-7",
abstract = "Chain-of-Thought (CoT) prompting enables complex reasoning in large language models (LLMs), including applications in information retrieval (IR). However, it often leads to overthinking, where models produce excessively long and semantically redundant traces with little or no benefit. We identify two key challenges in IR: redundant trajectories that revisit similar states and misguided reasoning that diverges from user intent. To address these, we propose State Machine Reasoning (SMR), a transition-based reasoning framework composed of discrete actions (REFINE, RERANK, STOP) that support early stopping and fine-grained control. Experiments on the BEIR and BRIGHT benchmarks show that improves retrieval performance (nDCG@10) by 3.4{\%} while reducing token usage by 74.4{\%}. It generalizes across LLMs and retrievers without requiring task-specific tuning, offering a practical alternative to conventional CoT reasoning."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lee-etal-2025-token">
<titleInfo>
<title>From Token to Action: State Machine Reasoning to Mitigate Overthinking in Information Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dohyeon</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yeonseok</namePart>
<namePart type="family">Jeong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seung-won</namePart>
<namePart type="family">Hwang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Chain-of-Thought (CoT) prompting enables complex reasoning in large language models (LLMs), including applications in information retrieval (IR). However, it often leads to overthinking, where models produce excessively long and semantically redundant traces with little or no benefit. We identify two key challenges in IR: redundant trajectories that revisit similar states and misguided reasoning that diverges from user intent. To address these, we propose State Machine Reasoning (SMR), a transition-based reasoning framework composed of discrete actions (REFINE, RERANK, STOP) that support early stopping and fine-grained control. Experiments on the BEIR and BRIGHT benchmarks show that improves retrieval performance (nDCG@10) by 3.4% while reducing token usage by 74.4%. It generalizes across LLMs and retrievers without requiring task-specific tuning, offering a practical alternative to conventional CoT reasoning.</abstract>
<identifier type="citekey">lee-etal-2025-token</identifier>
<identifier type="doi">10.18653/v1/2025.findings-emnlp.371</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.371/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>7048</start>
<end>7064</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T From Token to Action: State Machine Reasoning to Mitigate Overthinking in Information Retrieval
%A Lee, Dohyeon
%A Jeong, Yeonseok
%A Hwang, Seung-won
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F lee-etal-2025-token
%X Chain-of-Thought (CoT) prompting enables complex reasoning in large language models (LLMs), including applications in information retrieval (IR). However, it often leads to overthinking, where models produce excessively long and semantically redundant traces with little or no benefit. We identify two key challenges in IR: redundant trajectories that revisit similar states and misguided reasoning that diverges from user intent. To address these, we propose State Machine Reasoning (SMR), a transition-based reasoning framework composed of discrete actions (REFINE, RERANK, STOP) that support early stopping and fine-grained control. Experiments on the BEIR and BRIGHT benchmarks show that improves retrieval performance (nDCG@10) by 3.4% while reducing token usage by 74.4%. It generalizes across LLMs and retrievers without requiring task-specific tuning, offering a practical alternative to conventional CoT reasoning.
%R 10.18653/v1/2025.findings-emnlp.371
%U https://aclanthology.org/2025.findings-emnlp.371/
%U https://doi.org/10.18653/v1/2025.findings-emnlp.371
%P 7048-7064
Markdown (Informal)
[From Token to Action: State Machine Reasoning to Mitigate Overthinking in Information Retrieval](https://aclanthology.org/2025.findings-emnlp.371/) (Lee et al., Findings 2025)
ACL